Evaluation of HEVC decoders (SW, Hybrid and HW)

[NikosD]

Welcome to this thread regarding the evaluation of HEVC decoders (HW,Hybrid and SW)

The latest update with new samples for evaluating pure fixed-function HW HEVC decoders is here (https://forum.doom9.org/showthread.php?p=1799099#post1799099) testing Kabylake HD 610, Skylake HD 530, nVidia Turing GTX 1660 Nvidia Pascal GTX 1050 Ti/1060, Nvidia Maxwell 3rd gen GTX 960 and AMD Polaris RX 470 - 01/03/2017

Added some info on VP9 acceleration support of browsers using Youtube on various hardware (including Polaris cards) here (https://forum.doom9.org/showthread.php?p=1813127#post1813127) - 22/07/2017

The first official confirmation of Skylake's (HD 530) 8K (!) HEVC 8bit decoding capability in pure HW decoding mode here (http://forum.doom9.org/showthread.php?p=1737321#post1737321) by Yups - 05/09/2015

Here (http://forum.doom9.org/showthread.php?p=1735714#post1735714) is the comparison between Skylake HD 530 and Nvidia GTX 960 - 24/08/2015

Probably world's first Skylake 4K HEVC 8bit (HD 530) decoding results here (http://forum.doom9.org/showthread.php?p=1735325#post1735325) and here (http://forum.doom9.org/showthread.php?p=1735329#post1735329) and here (http://forum.doom9.org/showthread.php?p=1735586#post1735586) by Yups: (looks like the world's fastest HW decoder for 8bit HEVC) - 22/08/2015

For updated Optimus results, take a look here (http://forum.doom9.org/showthread.php?p=1724055#post1724055) - 27/05/2015


For those interested in hybrid only results on Intel's iGPU (DXVA, OpenCL), look at the results of the new PowerDVD 15 x86 (DXVA, OpenCL) and latest LAV 0.65/MPC 1.4.4.286 DXVA x64/x86 decoders here (http://forum.doom9.org/showthread.php?p=1718243#post1718243)- 22/04/2015

New updated CPU results for latest Lentoid v2.1.0.0, LAV v0.67.3 and MS H.265 decode (build 10586) here (http://forum.doom9.org/showthread.php?p=1748100#post1748100) - 29/11/2015

For those interested in CPU only results, look at the results of the new Lentoid x64/x86 v2.0.3.2, PowerDVD 15 on various CPU architectures and systems including Win 10 Build 10049 using latest LAV 0.65/MPC 1.4.4.286 decoders.
Also added a 10bit HEVC video here (http://forum.doom9.org/showthread.php?p=1717938#post1717938) - 22/04/2015


Added new results for the first public release of UHDcode, the HEVC decoder of MultiCoreWare here (http://forum.doom9.org/showthread.php?p=1715668#post1715668)- 01/04/2015


For HW decoder results of Nvidia 960 GTX, you have to see the post from Nevcairiel here (http://forum.doom9.org/showthread.php?p=1706902#post1706902) - 23/01/2015
(these are the world's first benchmark results of the fixed-function HEVC decoder inside GTX 960)

For a little older tests - 14/01/2015 - including PowerDVD v14 decoder, Microsoft's MFT Windows 10 decoder, a 10bit clip, on various CPU architectures (from Core 2 Duo up to Core i7 Haswell) take a look here (http://forum.doom9.org/showthread.php?p=1705352#post1705352)


The tool I use is always DXVA Checker you can download from here (http://bluesky23.yukishigure.com/en/index.html)


The first post (23/09/2014) starts here:

I used latest DXVA Checker x86/x64 version v3.1.2 using DXVA processing and a scaling resolution of 1280x720 for all clips.


1080p clips

1.ProRes-1080p@30fps-2Mbps
https://www.sendspace.com/file/mo8exg


LAV x64 i7-4790 158/230/251

LAV x86 i7-4790 121/151/188

LAV x64 i7-4790 (DXVA) 83/134/157

LAV x86 i7-4790 (DXVA) 84/133/158

Strongene x86 i7-4790 (OpenCL) 83/112/146

Strongene x86 i7-4790 97/109/111

LAV x64 i5-4200M 76/98/114

LAV x64 i5-4200M (DXVA) 60/84/96

LAV x64 Core 2 Quad 45/52/54




2.Elephants Dream-1080p@24fps-1.7Mbps
http://www.libde265.org/hevc-bitstreams/elephants-dream-1080-cfg02.mkv


LAV x64 i7-4790 93/195/242

LAV x64 i7-4790 (DXVA) 78/138/204

LAV x86 i7-4790 (DXVA) 66/135/200

LAV x86 i7-4790 62/133/259

Strongene x86 i7-4790 66/107/116

LAV x64 i5-4200M 40/97/130

LAV x64 i5-4200M (DXVA) 44/84/112

LAV x64 Core 2 Quad 35/52/55

Strongene x86 i7-4790 (OpenCL) Crashed



3.Big Buck Bunny-1080p@60fps-2Mbps
http://www.libde265.org/hevc-bitstreams/bbb-1920x1080-cfg06.mkv


LAV x64 i7-4790 144/232/267

LAV x64 i7-4790 (DXVA) 123/165/230

LAV x86 i7-4790 (DXVA) 113/161/201

LAV x86 i7-4790 67/146/259

LAV x64 i5-4200M 66/109/131

LAV x64 i5-4200M (DXVA) 68/97/114

LAV x64 Core 2 Quad 45/53/55

Strongene x86 i7-4790 (OpenCL) Crashed

Strongene x86 i7-4790 Crashed just before the end of the file



UHD - 3840x2160p clips


1.Beauty-2160p@30fps-12.3Mbps
http://ultravideo.cs.tut.fi/video/Beauty_3840x2160_120fps_420_8bit_HEVC_MP4.mp4


LAV x64 i7-4790 51/63/65

Strongene x86 i7-4790 (OpenCL) 26/47/47

LAV x86 i7-479025/40/43

Strongene x86 i7-4790 20/34/35

LAV x64 i7-4790 (DXVA) 29/33/35

LAV x86 i7-4790 (DXVA) 30/32/33

LAV x64 i5-4200M 15/25/27

LAV x64 i5-4200M (DXVA) 19/21/22

LAV x64 Core 2 Quad 6/13/13



2.Fitness-2160p@30fps-8Mbps
http://cloud.ultrahdtv.net/fitness-trailer-8000.mkv


LAV x64 i7-4790 57/71/82

LAV x64 i7-4790 (DXVA) 41/55/77

LAV x86 i7-4790 (DXVA) 40/52/76

LAV x86 i7-4790 35/48/71

Strongene x86 i7-4790 28/40/50

LAV x64 i5-4200M (DXVA) 25/30/39

LAV x64 i5-4200M 22/29/38

LAV x64 Core 2 Quad 9/13/13

Strongene x86 i7-4790 (OpenCL) Crashed



3.Ducks-2160p@50fps-4Mbps
https://www.sendspace.com/file/cyiv49


LAV x64 i7-4790 62/72/73

Strongene x86 i7-4790 (OpenCL) 29/46/50

LAV x86 i7-4790 (DXVA) 43/45/47

LAV x64 i7-4790 (DXVA) 42/45/47

Strongene x86 i7-4790 27/41/42

LAV x86 i7-4790 29/40/44

LAV x64 i5-4200M 25/31/33

LAV x64 i5-4200M (DXVA) 27/29/30

LAV x64 Core 2 Quad 10/13/13



Comments:

I used latest nightly of LAV filters 0.62.46 (24-09-2014) with LAV Video threads set to Auto and latest Strongene's OpenCL and CPU HEVC decoders.

The Core i7 system tested was:
Win 8.1 Pro x64 - Core i7-4790 - HD 4600 - Drivers v.3907

The Core i5 system tested was:
Win 8.1 x64 - Core i5-4200M@2.5GHz (battery mode) - HD 4600 - Drivers v.3907 - Nvidia 740M - Drivers 344.11 - Optimus technology

The Core 2 Quad system tested was:
Win 8.1 Pro x64 - Core 2 Quad Q9550@2.26GHz (266MHz x 8.5) - Nvidia GT610 - Drivers 344.11

For laptop and Nvidia hybrid decoder, see my post here (http://forum.doom9.org/showthread.php?p=1694992#post1694992)

For tests using more threads than Auto, null renderer, overclocked CPUs and different OS (Win 7) check this (http://forum.doom9.org/showthread.php?p=1695007#post1695007) post:

CPU decoders for Core i7


The CPU decoders were using GPU as low as possible ~25%-35% with 600MHz clock, but Strongene's CPU was a little higher ~80%@600MHz

1080p
For LAV x86/x64 CPU decoders, CPU utilization was 55% - 60% using all 8 threads with a clock of 3.8GHz, for Strongene CPU decoder it was only 16% !!! with a clock of ~4.0GHz with only a few threads enabled.

2160p
For LAV x64 CPU decoder, CPU utilization was 77% - 91% using all 8 threads with a clock of 3.8GHz, for LAV x86 CPU CPU utilization was 92% - 94% using all 8 threads with a clock of 3.8GHz, for Strongene CPU decoder it was only 25% -29% !!! with a clock of ~4.0GHz with only a few threads enabled.



CPU decoders for Core i5/ Core 2 Quad


I used only LAV x64 decoder and the results are completely weird.

Core i5-4200M@2.5GHz is a dual core processor with four threads (2C/4T) and achieved a CPU utilization of 76% - 86% for 1080p and 76% - 89% for 2160p

Core 2 Quad Q9550@2.26GHz is a quad core processor with four threads (4C/4T) and achieved a CPU utilization of 50% - 58% for 1080p and 55% - 60% for 2160p

For all clips the performance of the dual core Core i5@2.5GHz is almost double than the quad Core 2 Quad@2.26GHz with more than 50% CPU utilization (!) for the dual core (2C/4T) chip (!!)



GPU decoders for Core i7


1080p
The DXVA decoders were using GPU at ~100% and max 1200MHz clock and CPU at ~14% with a clock of 2.6GHz - 4.0GHz

Strongene's OpenCL decoder used GPU like Strongene's CPU decoder - only a few MHz higher (600MHz - 750MHz) and a CPU usage of ~11% at 4.0GHz


2160p
The DXVA decoders were using GPU at ~90% and max 1200MHz clock and CPU at ~14% with a clock of 2.6GHz - 4.0GHz.

The memory usage went to the limit of ~960MB-990MB, the maximum that iGPU can use.

Strongene's OpenCL decoder used GPU a lot more than 1080p at 85%@850MHz - 900MHz and the CPU usage of 36% - 50% at 3.8GHz-4.0GHz which is double than Strongene's CPU decoder!


GPU decoders for Core-i5

See here (http://forum.doom9.org/showthread.php?p=1694992#post1694992)



----------------------------------------------------------------------------------------------------------------------------------
LAV DXVA x86 and x64 have almost the same performance and in 2160p they are slower than OpenCL decoder, mainly because of the CPU usage of OpenCL decoder and not GPU.

It's interesting that LAV DXVA x86/x64 is faster than LAV CPU x86 in 1080p clips and low bitrate 2160p clips.
But in just 12Mbps 2160p clip, the LAV x86 CPU is faster than DXVA.

During benchmarking LAV CPU decoders were using more power (watt) than DXVA or OpenCL decoder, but I think during playback the opposite occurs (I haven't tested yet)

I used latest Intel's GPA tool to monitor any QuickSync (fixed-function HW) activity, but it was a dead zero during DXVA decoding.

So everything is decoded in EUs and a little CPU for HEVC DXVA2 decoder.

----------------------------------------------------------------------------------------------------------------------------------

Conclusion

LAV DXVA uses maximum iGPU memory of 1GB for 4K decoding and with a low bandwidth of even 12Mbps is slower even than LAV x86 CPU.
It's more useful for lower resolutions.

Strongene's OpenCL decoder is definitely a lot more useful for 4K than 1080p, but it has incompatibilities and in order to be fast at 4K it uses a lot the CPU, not so much the GPU.

Strongene's CPU decoder is almost always slower of all.

LAV x64 CPU is always faster than anything else in all bitrates and resolutions.

CPU utilization of all CPU decoders, rises with the increased resolution and low bitrate.
Best example is Duck 2160p@4Mbps clip with LAV x64 CPU utilization of 91% and LAV x86 at 94%

When bandwidth increases a little, the CPU utilization and decoding performance drops more.

For all decoders I used a good case scenario of a 1280x720 display resolution.

When native resolution is used for display at 1080p and 2160p, the results are lower by a good percentage.

For 4K BluRay with a bitrate of 100Mbps and 10 bit resolution, expect CPU decoders and hybrid decoders to be useless (?)* even with Haswell-E or Xeon processors.

We are going to need definitely pure fixed-function HW decoders for 4K BluRay.

On the other hand, 4K BluRay will appear on winter holidays of 2015, so until then, CPU and hybrid decoders are just fine for low bandwidth and low fps clips that HEVC is the best codec to use.

I have already encoded 4K H.264 and 4K H.265 up to 600Mbps (!) and I can say for sure that 4K H.264 performance of Haswell QuickSync decoder is about 190fps for 4K H.264 100Mbps clip at 1280x720 display resolution.

Intel and Nvidia decided to offer a hybrid solution now, which is a useful choice for low fps, low bandwidth clips even at 4K resolution and by the time of 4K BluRay arrival, fixed-function decoders will be ready.

Hopefully AMD will follow.

* according to these (http://forum.doom9.org/showthread.php?p=1694854#post1694854) results by cyberbeing an overclocked CPU - even without HT - and a real fast Nvidia card can handle even more difficult to decode 4K HEVC clips at greater speed than my results.

But still 4K BluRay will be out of their reach I think.

[NikosD]

Next thing to test is a special driver from Nvidia or Intel or a LAV filters special version with significant gain in performance (CPU or Hybrid) for Nvidia or Intel.

I'll definitely try to repeat the tests with Optimus laptop to check if anything has changed regarding Nvidia's hybrid decoder performance.

If any of you has interesting samples (difficult to decode), I would give them a try, especially in 1080p resolution for which I didn't find a lot of samples.

[wanezhiling]

Off topic: Today's PotPlayer 1.6.49903 has supported HEVC CUDA/DXVA decoding as well. :)

http://i.imgur.com/hcLDgBG.png

[nevcairiel]

They are fast with stealing my code, apparently.

[sneaker_ger]

If any of you has interesting samples (difficult to decode), I would give them a try

Astra 4K demo channel:
https://www.dropbox.com/s/6rbqlrrhhvbz79l/Astra%20UHD.ts?dl=0 (probably not cut correctly at GOP border)

But it uses 10 bit coding - I don't know if Nvidia and Strongene support that?

[JohnLai]

They are fast with stealing my code, apparently.

So, what is the action you will take against them?

This code stealing activity is frowned upon.

[NikosD]

They are fast with stealing my code, apparently.

So, what is the action you will take against them?

This code stealing activity is frowned upon.

Guys, please don't start this discussion here.

I think there are threads here in Doom9 with that subject - I think it has been discussed thoroughly - or you can start a new thread about that.

Astra 4K demo channel:
https://www.dropbox.com/s/6rbqlrrhhvbz79l/Astra%20UHD.ts?dl=0 (probably not cut correctly at GOP border)

But it uses 10 bit coding - I don't know if Nvidia and Strongene support that?


Well, you are going to make me buy a new CPU for that file!

I used latest nightly MPC-HC x64 and I got a lot of stuttering using EVR-CP.
I had to use plain EVR in order to be smoother, but I can't say it was absolutely smooth.

LAV DXVA, Strongene's OpenCL and CPU decoders are all incompatible with 10-bit HEVC, so I can't use them for decoding comparisons.

I don't know yet about Nvidia's HEVC decoder, but I think only LAV CPU can decode 10-bit HEVC clips.

BTW, I have a lot of 10 bit 4K HEVC clips from here http://demo-uhd3d.com/ which all of them stutter during playback.

It's not exactly stutter or judder, it's like repeatedly spasmodic movement, after a few seconds of normal playback each time - you can call it stutter.

The problem with those 6 demo files I have, is that the CPU utilization is below 50% when the movement looks like stuttered.

I thought the problem was in the encoding or muxing and stopped downloading other clips from that site for that reason.

Is it possible the decoding speed to be the reason of stuttering ?

Those clips are huge.

6 of them are 5.15GBytes on my disk.

I don't know if you grabbed your sample from there and you cut a smaller sample, but if you have a link of the original source I would like to download the whole file.

Also I would be interested in HEVC 1080p files too.

[sneaker_ger]

Well, you are going to make me buy a new CPU for that file!
Yes, it's very demanding. My core i7-860 does less than the at least required 50 fps on that sample even with Null Renderer. But compared to the 4K Blu-Ray this is low bitrate. We're looking at 4Kp60, 10 bit and more bitrate than current Blu-Rays.

I don't know if you grabbed your sample from there and you cut a smaller sample, but if you have a link of the original source I would like to download the whole file.
It's a direct capture from the Astra 19.2° satellite.

[NikosD]

Yes, it's very demanding. My core i7-860 does less than the at least required 50 fps on that sample even with Null Renderer.



I get 56/67/77 with Null Renderer (DXVA Decoding using DXVA Checker) using 8 threads @ 3.8GHz with 76% utilization.


But compared to the 4K Blu-Ray this is low bitrate. We're looking at 4Kp60, 10 bit and more bitrate than current Blu-Rays.


By that time, I think around winter holidays of 2015, we will have HW decoders which will eat 4K Bluray for breakfast :)


It's a direct capture from the Astra 19.2° satellite.


OK, thanks.

[huhn]

the potplayer hevc hardware decoder is used with 10 bit hevc it totally fails and uses only 1 thread for decoding but there is something like a picture.

the output is reported as nv12 by madVR

[NikosD]

Added 2160p clips and changed a lot the comments for the new results.

I added a conclusion, too.

[cyberbeing]

i5-3570K @4.4Ghz, 16GB DDR3 1866Mhz 8-9-9-24
NVIDIA GTX 770 2GB, PCI-E 3.0 x8
Win7 SP1 x64
LAV Filters git-33f0b3 (2014-09-21)
DXVA Checker 3.1.2



1.Beauty-2160p@30fps-12.3Mbps

LAV x64 CPU (Null) 91/106/110
LAV x64 CPU (1280x720 scale) 78/100/110

LAV x64 DXVA (Null) 78/86/87
LAV x64 DXVA (1280x720 scale) 78/84/86

LAV x86 DXVA (Null) 76/83/86

LAV x86 CPU (Null) 36/47/53


2.Fitness-2160p@30fps-8Mbps

LAV x64 CPU (Null) 109/132/142
LAV x64 CPU (1280x720 scale) 88/125/142

LAV x64 DXVA (Null) 100/113/146
LAV x64 DXVA (1280x720 scale) 97/111/125

LAV x86 DXVA (Null) 86/101/123

LAV x86 CPU (Null) 41/62/118


3.Ducks-2160p@50fps-4Mbps

LAV x64 DXVA (Null) 193/197/198
LAV x64 DXVA (1280x720 scale) 183/187/189

LAV x86 DXVA (Null) 185/189/190

LAV x64 CPU (Null) 128/140/144
LAV x64 CPU (1280x720 scale) 107/131/140

LAV x86 CPU (Null) 36/48/54


Astra-UHD@50fps-18Mbps (10bit)
https://www.dropbox.com/s/6rbqlrrhhvbz79l/Astra%20UHD.ts?dl=0

LAV x64 CPU (Null-P010) 70/86/107
LAV x64 CPU (1280x720 scale-NV12) 65/86/107

LAV x86 CPU (Null-P010) 28/44/86


UHD_ENT_Transformer_Quad@24fps-51Mbps (10bit)
http://demo-uhd3d.com/files/4k/Transformers_extinction_Trailer_4K.zip

LAV x64 CPU (Null-P010) 45/59/102
LAV x64 CPU (1280x720 scale-NV12) 45/59/102

LAV x86 CPU (Null-P010) 25/39/86

[nevcairiel]

Are you sure DXVA was actually used? The performance seems extremely high for 2160p content. I never saw it go above 50 fps at that resolution, no matter how "simple" i made the encodes.

[cyberbeing]

Yes, I'm sure 'DXVA (Native)' was used. CPU usage was low, GPU usage was high. LAV reported 'dxva2n' during playback/benchmark.

Ducks UHD, GPU load ~90%, GTX 770 Core 1200Mhz (Boost)
Fitness UHD, GPU load ~55%, GTX 770 Core 1110Mhz (non-Boost)
Beauty UHD, GPU load ~65%, GTX 770 Core 1110Mhz (non-Boost)

[NikosD]

Yes, I'm sure 'DXVA (Native)' was used. CPU usage was low, GPU usage was high. LAV reported 'dxva2n' during playback/benchmark.

Ducks UHD, GPU load ~90%, GTX 770 Core 1200Mhz (Boost)
Fitness UHD, GPU load ~55%, GTX 770 Core 1110Mhz (non-Boost)
Beauty UHD, GPU load ~65%, GTX 770 Core 1110Mhz (non-Boost)

You have a system with an HD 4600 card and a discrete Nvidia card.

I think HEVC DXVA2 decoder is only for Intel cards, tomorrow I'll try Nvidia cards to see if it works for them too.

I'm sure that Nvidia cards have NVCUVID ability to decode HEVC.

The results of LAV DXVA for what card are referred to ?

Is it Nvidia using NVCUVID ?

[huhn]

no nvidia HEVC DXVA2 works with better kampler cards.

[cyberbeing]

I don't even have the driver from my CPU's HD4000 installed and it's disabled in the BIOS, so it's impossible that it would be used.

LAV added support for both DXVA2 & CUVID on supported NVIDIA cards (Kepler and newer?).

Use of the CUVID or DXVA2 Copyback modes is of course much slower than DXVA2 Native, especially since I'm only using a PCI-E 3.0 x8 link. Though GPU & CPU usage is also much lower when benching in these modes, since the hybrid decoder seems to become heavily bottlenecked by 3840x2160 copyback performance.

For example:

Ducks-2160p@50fps-4Mbps
LAV x64 DXVA Native (Null) 193/197/198
LAV x64 CUVID (Null) 73/78/78
LAV x64 DXVA Copyback (Null) 69/73/73

Fitness-2160p@30fps-8Mbps
LAV x64 DXVA Native (Null) 100/113/146
LAV x64 CUVID (Null) 50/57/70
LAV x64 DXVA Copyback (Null) 49/55/66

Beauty-2160p@30fps-12.3Mbps
LAV x64 DXVA Native (Null) 78/86/87
LAV x64 CUVID (Null) 43/51/55
LAV x64 DXVA Copyback (Null) 41/48/50

[NikosD]

Nice!

I didn't know.

So I have much work to do tomorrow if I want to test every possible decoder for 740M.

Hopefully I will see HEVC DXVA2 for 740M.

Yes your CPU is Ivy so your card is HD 4000, not 4600.

I thought that CUVID due to 3D clocks and zero memory performance penalty would be about equal to DXVA2 native.

Thanks for the info!

[huhn]

CUVID and DXVA2 use the same decoder so CUVID is like DXVA2 copyback at least with new cards.
currently i don't see any benefit with using CUVID over DXVA2 copyback or native

[nevcairiel]

I think I mostly tested performance with Copy-Back, so that may explain it, but I'm surprised its such a massive difference, wonder if I can tune something there..
On H.264 for example, the difference was always minimal, but i guess that was 1080p and not 2160p.

[NikosD]

CUVID and DXVA2 use the same decoder so CUVID is like DXVA2 copyback at least with new cards.
currently i don't see any benefit with using CUVID over DXVA2 copyback or native

From the words of nevcairiel, I was under the impression that Nvidia's hybrid HEVC decoder was a CUVID only feature, like MPEG4-ASP

[nevcairiel]

Who ever said MPEG4-ASP is a CUVID only feature? Its just that noone ever bothered to implement it in DXVA2, because its quite a bit of work for practically no return value. :)
With CUVID, it just comes for free, since its all handled in the driver, not much code needed at all.

HEVC is obviously implemented in DXVA2, and as such it also works on NVIDIA of course, and not only through CUVID.

[NikosD]

OK I'll test it and I'll post the results.

[sheppaul]

There is a MPEG4-ASP DxVA decoder in potplayer though.

I believed the code was from lav.

[huhn]

There is a MPEG4-ASP DxVA decoder in potplayer though.

I believed the code was from lav.

why do you think so there is no DXVA decoder for ASP in lavfilter only CUVID.

[edison]

The x64 LAV video does good job here, but there is another problem, we still have not a mature replay suit. for example: there is not a x64 DTS-HD software audio decoder so far.
So, we still need to wait for a real HW decoder which support 4K 60p(at least) HEVC( GM1XX does not have a full HW HEVC decoder, but GM2XX does).

[huhn]

The x64 LAV video does good job here, but there is another problem, we still have not a mature replay suit. for example: there is not a x64 DTS-HD software audio decoder so far.
So, we still need to wait for a real HW decoder which support 4K 60p(at least) HEVC( GM1XX does not have a full HW HEVC decoder, but GM2XX does).

the new nvidia 980/970 didn't have a full HW decoder for HEVC they only have a HW HEVC encoder nvenc used for shadow play of cause they don't use HEVC encoding at the current version but the card can do it.

you can read about this here.

http://www.anandtech.com/show/8526/nvidia-geforce-gtx-980-review/5

[sheppaul]

why do you think so there is no DXVA decoder for ASP in lavfilter only CUVID.

There is a checkbox for that though it is disabled by default.

I didn't know that it is not implemented.

BTW, the hybrid hevc decoder is pretty disappointing. Is there any room to improve?

[videoh]

BTW, the hybrid hevc decoder is pretty disappointing. In what way do you find it to be disappointing?

[NikosD]

BTW, the hybrid hevc decoder is pretty disappointing. Is there any room to improve?

From the results of cyberbeing using Nvidia 770, the hybrid decoder using DXVAn can be a lot faster than CPU decoder on my CPU Core i7-4790 and 2160p clips.

Look at the Ducks sample.
It's more than two times faster.

[NikosD]

Using a laptop with Nvidia's Optimus technology proved a difficult task for video decoding benchmarking.

First of all, I didn't find a way to disable it in BIOS, so I had to live with it.

Using initially Intel's HD 4600 iGPU inside the Core i5-4200M, I saw that even when DXVA native was used during video decoding, the HD 4600 never reached max clocks with maximum utilization.

The clocks were most of the time at 850MHz@100% usage for 1080p and 2160p clips with a max clock of 1150MHz.

Also, the CPU utilization was too high for DXVA native mode, ~40% for 1080p clips and 33% - 57% for 2160p and the CPU clock went to max 2.5GHz a lot of times during benchmarking.

It looks like a system with an iGPU and a discrete Nvidia card on the same system using Optimus technology, can't utilize perfectly DXVA native mode for both GPUs.

Of course DXVA copy-back was even slower.

From the Control Panel I chose max performance for iGPU.
Also I tried using the laptop in battery and plugged-in.

Nothing changed.



But with Nvidia's 740M GPU the problems were a lot bigger.

Starting with 340.52 driver, the clocks went high at 980MHz (max 1060MHz) but with low GPU utilization.
Then the video dropped to 0fps and stopped working.

Unfortunately that happened with almost all clips I tried and all modes (DXVA copy-back, NVCUVID)

I tried both LAV x86 and x64 with no success.

I did an update to the latest driver 344.11, but still had issues.

The GPU clocks went further down to 140MHz - 230MHz (!) with very little GPU usage, but at least most of the clips were fully decoded without sudden stops, but not all the clips I tried.
Some, still have incompatibilities.

So, there are still compatibility problems with 740M even with latest drivers and LAV nightly 0.62.46

The results were extremely low for the clips that decoded fine and some didn't reach the end, so I decided to not include any results.

I don't know if it's Optimus technology problem or driver or LAV Video.

I will try again after a while, to see if anything has changed.

[huhn]

the max gpu clock is heat controlled for both nvidia and intel.

and this is a laptop they are not know for efficient heat management.

[NikosD]

the max gpu clock is heat controlled for both nvidia and intel.

and this is a laptop they are not know for efficient heat management.

I was way below the heat throttling of both GPUs.

Especially for Nvidia, the issue is certainly not there.


i5-3570K @4.4Ghz, 16GB DDR3 1866Mhz 8-9-9-24
NVIDIA GTX 770 2GB, PCI-E 3.0 x8
Win7 SP1 x64
LAV Filters git-33f0b3 (2014-09-21)
DXVA Checker 3.1.2



OK you have a:
Core i5-3570K @4.4Ghz (Ivy) 4cores/4threads (No HT) - Win7 SP1 x64 - 16GB - DDR3 1866MHz CL9 - LAV Filters git-33f0b3 (2014-09-21) - DXVA Checker 3.1.2

I have a:
Core i7-4790@3.8GHz (Haswell) 4cores/8threads (HT On) - Win 8.1 Pro x64 - 8GB - DDR3 1600MHz CL9 - LAV filters git 52 (25-09-2014) - DXVA Checker 3.1.2 (DXVA Decoding)

Your processor has 16% higher frequency, but it's only 4 threads vs 8 threads and it's older architecture (Ivy vs Haswell)


Results:


1.Beauty-2160p@30fps-12.3Mbps


LAV x64 i5-3570K 97/107/111 CPU Utilization: 95% (Threads=12) (Windows 7) (4.4GHz)

LAV x64 i5-3570K 89/103/109 CPU Utilization: 91% (Threads=12) (Windows 8.1) (4.4GHz)

LAV x64 i7-3770K 85/101/105 CPU Utilization: 79% (Threads=16) (HT on) (Windows 7) (4.0GHz)

LAV x64 i7-4790 82/95/99 CPU Utilisation: 78% (Threads=16) (HT on) (Windows 8.1) (3.8GHz)

LAV x64 i7-3770K 88/93/100 CPU Utilization: 95% (Threads=16) (HT off) (Windows 7) (4.0GHz)

LAV x64 i7-3770K 80/93/99 CPU Utilization: 70% (Thread=Auto) (HT on) (Windows 7) (4.0GHz)

LAV x64 i7-3770K 77/92/97 CPU Utilization: 94% (Threads=12) (HT off) (Windows 7) (4.0GHz)

LAV x64 i7-4790 78/90/95 CPU Utilisation: 73% (Threads=Auto) (HT on) (Windows 8.1) (3.8GHz)

LAV x64 i5-3570K 76/82/86 CPU Utilization: 72% (Threads=Auto) (Windows 7) (4.4GHz)

LAV x64 i7-4790 69/81/87 CPU Utilisation: 90% (Threads=16) (HT off) (Windows 8.1) (3.8GHz)

LAV x64 i7-4790 69/81/85 CPU Utilisation: 90% (Threads=12) (HT off) (Windows 8.1) (3.8GHz)

LAV x64 i7-3770K 61/71/76 CPU Utilization: 71% (Threads=Auto) (HT off) (Windows 7) (4.0GHz)

LAV x64 i7-4790 61/68/72 CPU Utilisation: 90% (Threads=Auto) (HT off) (Windows 8.1) (3.8GHz)





2.Fitness-2160p@30fps-8Mbps

LAV x64 i5-3570K (Null) 109/132/142 Threads=12
LAV x64 i7-4790 (Null) 99/120/148 CPU Utilisation: 92% Threads=16
LAV x64 i7-4790 (Null) 97/113/136 CPU Utilisation: 85% Threads=Auto



3.Ducks-2160p@50fps-4Mbps

LAV x64 i5-3570K (Null) 128/140/144 Threads=12
LAV x64 i7-4790 (Null) 109/122/127 CPU Utilisation: 92% Threads=16
LAV x64 i7-4790 (Null) 110/120/122 CPU Utilisation: 91% Threads=Auto



Astra-UHD@50fps-18Mbps (10bit)


LAV x64 i5-3570K 71/86/108 CPU Utilization: 94% (Threads=12) (Windows 7) (4.4GHz)

LAV x64 i5-3570K 76/82/86 CPU Utilization: 72% (Threads=Auto) (Windows 7) (4.4GHz)

LAV x64 i5-3570K 73/81/86 CPU Utilization: 69% (Threads=Auto) (Windows 8.1) (4.4GHz)

LAV x64 i7-4790 61/72/85 CPU Utilisation: 80% (Threads=16) (HT on) (Windows 8.1) (3.8GHz)

LAV x64 i7-4790 59/69/81 CPU Utilisation: 76% (Threads=Auto) (HT on) (Windows 8.1) (3.8GHz)

LAV x64 i5-3570K 55/69/95 CPU Utilization: 75% (Threads=Auto) (Windows 7) (4.4GHz)

LAV x64 i7-4790 51/64/77 CPU Utilisation: 93% (Threads=12) (HT off) (Windows 8.1) (3.8GHz)

LAV x64 i7-4790 47/63/77 CPU Utilisation: 92% (Threads=16) (HT off) (Windows 8.1) (3.8GHz)

LAV x64 i7-4790 47/55/76 CPU Utilisation: 79% (Threads=Auto) (HT off) (Windows 8.1) (3.8GHz)



UHD_ENT_Transformer_Quad@24fps-51Mbps (10bit)

LAV x64 i5-3570K(Null-P010) 45/59/102
LAV x64 i7-4790 (Null) 45/58/85 CPU Utilisation: 96% Threads=16
LAV x64 i7-4790 (Null-P010) 45/57/80 CPU Utilisation: 94% Threads=Auto


Can you explain me at least the last result, a Core i7-4790 with a CPU utilization of 96% (full 8 threads decoding) to have less decoding performance than a Core i5-3570K with 15% higher frequency.

I find it unbelievable.

I used DXVA Checker's v3.1.2 DXVA decoding choice for Null renderer and latest (today's) LAV filters 0.62.52

Can you try that tool with that filters version to check the performance reporting CPU utilization too?

Thanks!

[cyberbeing]

All my previous results were with LAV Video set to Threads=12, since as you also noticed, CPU utilization and resulting performance is lower than expected on some HEVC samples with LAV set to Threads=Auto. I'd assume this is the discrepancy you are seeing.

[NikosD]

All my previous results were with LAV Video set to Threads=12, since as you also noticed, CPU utilization and resulting performance is lower than expected on some HEVC samples with LAV set to Threads=Auto. I'd assume this is the discrepancy you are seeing.

OK.

I tried first putting Threads=12, but it was equal or a little slower than Auto for my CPU.

So I went all the way up to Threads=16.

I edited my previous post to show the results of Threads=16

There is an increase, but still I can't reach you.

I still find it extremely weird.

Do you have a secret ? ;)

Did you use GraphStudioNext for benchmarking Null renderer ? :p

I can't figure it out, unless it's the different LAV filters version.

I can't think of anything else.

[cyberbeing]

I'm not doing anything special:
DXVAChecker 3.1.2 -> Decoder tab -> Drag/Drop Video -> Click Arrow -> Benchmark -> DXVA decoding

Threading differences between Win7 & Win8, HT vs no-HT, architecture jumps not always being superior in all metrics, and that Intel CPUs sometimes unbottleneck themselves when overclocked could all be possible explanations. Also since you have a laptop, it's possible your CPU is power/thermal throttling and not using maximum TurboBoost when benchmarking. If you download RealTemp TI (http://forum.techinferno.com/downloads.php?do=file&id=43), it should show you the actual clock speed at any given time.


It looks like LAV HEVC threading problem I was referring to only occurs on the following two samples:

3570K@4.4Ghz
LAV Filters git-1d591 (Betaking build)

1.Beauty-2160p@30fps-12.3Mbps

LAV x64 i5-3570K 97/107/111 CPU Utilization: 95% (Threads=12)
LAV x64 i5-3570K 76/82/86 CPU Utilization: 72% (Threads=Auto)


Astra-UHD@50fps-18Mbps (10bit)

LAV x64 i5-3570K 71/86/108 CPU Utilization: 94% (Threads=12)
LAV x64 i5-3570K 55/69/95 CPU Utilization: 75% (Threads=Auto)

The other samples had good utilization (>92%) with Threads=Auto, and only ~1% lower performance vs Threads=12. Not so for the two samples above.

[huhn]

@cyberbeing
he uses processing with 1280x720.

my i7 3770k 4.0 ghz

get's this with 62.52 and processing 1280x720

Beauty-2160p@30fps

12 threads CPU 64 83 90 / 60 % cpu usage

[cyberbeing]

@cyberbeing
he uses processing with 1280x720.

Above (http://forum.doom9.org/showpost.php?p=1695007&postcount=34) NikosD was testing Null, which implies no scaling.

My results (http://forum.doom9.org/showpost.php?p=1694854&postcount=13) using 'DXVA Processing' and 1280x720 scaling with actual renderer output, explicitly state so. There is not much difference in performance either way.

Since you both have CPUs with Hyperthreading, try disabling HT and see if your results improve. And you huhn, since you have a 3770k, could try matching my 4.4Ghz overclock. If you are only seeing 60% CPU Utilization on your 3770K that is extremely strange...were you also using Win8 and not Win7 like I am? I'll try booting into Win8.1 later and see if there is any difference.

Also, the LAV Filters nightly builds I've been using for testing are those from Betaking: here (http://pan.baidu.com/s/1gd1its3#dir/path=%2F%E8%87%AA%E7%BC%96%E8%AF%91%E8%BD%AF%E4%BB%B6). Which builds have you guys been using?

Edit (Win7 vs Win8.1 Threading):
Windows 8.1 does seem to have marginally lower CPU Utilization and benchmark performance compared to Windows 7, but not significant enough to explain what NikosD or huhn are seeing.

1.Beauty-2160p@30fps-12.3Mbps
LAV x64 i5-3570K 97/107/111 CPU Utilization: 95% (Threads=12) (Windows 7)
LAV x64 i5-3570K 89/103/109 CPU Utilization: 91% (Threads=12) (Windows 8.1)

LAV x64 i5-3570K 76/82/86 CPU Utilization: 72% (Threads=Auto) (Windows 7)
LAV x64 i5-3570K 73/81/86 CPU Utilization: 69% (Threads=Auto) (Windows 8.1)

Edit2 (Betaking 2014-09-25 vs K-Lite 10.76 LAV nightly build):
No difference in performance.

[huhn]

i used windows 7.

lavfilter is 0.62.0.52 i took it from KLCP

4.4 GHz results in a blue screen after reboot. my 3770k has pretty bad quality silicon so i stay at 4.0.
my RAM was at 1333 for some reasons it now at normal 1600.
the system wasn't rebooted for some time looks a lot better with 1600 and rebooted system:

Beauty_3840x2160_120fps_420_8bit_HEVC_MP4

80 93 99 / 70% threads auto (should be 12 with 8 thread system)
85 101 105 / 79% threads 16

i'm pretty sure an i7 will look pretty good with 24-32 threads.

all used Null rendering

edit:
non HT

61 71 76 / 71% threads auto (should be 6)
77 92 97 / 94% threads 12
88 93 100 / 95% threads 16


looks like using 3 times the number of threads the CPU got helps for HEVC decoding.

12 thread with core 4 cpu and 12 threads with 4 core cpu with HT threads have about the same speed the min FPS is a lot higher with HT.
at 16 threads it looks like HT give a decent speed boost where non HT has no real difference except min fps.
i guess HT does a good job with a lot of threads too but this will take "a lot" of RAM. i guess even with 32 threads this shouldn't be a real problem for people with 8 thread CPUs.

[nevcairiel]

Some clips are just not encoded in a way that would be beneficial to multi-threading, and while you can try throwing more and more threads at it to try to improve the CPU utilization, its usually a bad solution.
For such clips it may be useful to combine both frame and slice multi-threading, so that big I frames can use slice threading (or wpp) to speed up their decoding, since everything hinges on them being ready as a reference for the others. I've seen preliminary patches for this, but its never been finished.

[huhn]

so on a file with only 1 I frame and a length of over 2000 frame this should be meaningless because only the first frame would benefit from throwing threads against it?

number with a i3 4130 at 3.4 ghz
37 48 54 / 75% threads auto (6)
38 55 62 / 95% threads 12

[nevcairiel]

other frames with a lot of complexity can also benefit, generally such bottlenecks start because its waiting for a reference frame, although a 2000 frame GOP is already a terribly encoded video.
Usually I frames are the most complex, since they are an order of magnitude bigger than any other frames.

[NikosD]

Astra-UHD@50fps-18Mbps (10bit)

LAV x64 i5-3570K 71/86/108 CPU Utilization: 94% (Threads=12)
LAV x64 i5-3570K 55/69/95 CPU Utilization: 75% (Threads=Auto)



LAV x64 i5-3570K 76/82/86 CPU Utilization: 72% (Threads=Auto) (Windows 7)
LAV x64 i5-3570K 73/81/86 CPU Utilization: 69% (Threads=Auto) (Windows 8.1)



The results for ASTRA clip in those two posts (one after the other), do not interpret.

[NikosD]

Also since you have a laptop, it's possible your CPU is power/thermal throttling and not using maximum TurboBoost when benchmarking. If you download RealTemp TI (http://forum.techinferno.com/downloads.php?do=file&id=43), it should show you the actual clock speed at any given time.


My Core i7-4790 results are from a desktop system.
I think there is no laptop with such a hot (!) CPU.
They all use the M or U branches.

The results of Core i5-4200M are from laptop.

I use CoreTemp for real turbo boosted frequencies and it's 3.8GHz all the time during benchmarking.


It looks like LAV HEVC threading problem I was referring to only occurs on the following two samples:

3570K@4.4Ghz
LAV Filters git-1d591 (Betaking build)

1.Beauty-2160p@30fps-12.3Mbps

LAV x64 i5-3570K 97/107/111 CPU Utilization: 95% (Threads=12)
LAV x64 i5-3570K 76/82/86 CPU Utilization: 72% (Threads=Auto)


Astra-UHD@50fps-18Mbps (10bit)

LAV x64 i5-3570K 71/86/108 CPU Utilization: 94% (Threads=12)
LAV x64 i5-3570K 55/69/95 CPU Utilization: 75% (Threads=Auto)

The other samples had good utilization (>92%) with Threads=Auto, and only ~1% lower performance vs Threads=12. Not so for the two samples above.

Exactly.
My tests show exactly this behavior, so the explanation of performance difference, still remains for the other clips which don't have benefits by raising the number of threads.

Update:

Looking more carefully the "refreshed" results of huhn using a tad faster 4.0GHz CPU, I think that Win 7 could probably matters most for the performance difference.


Since you both have CPUs with Hyperthreading, try disabling HT and see if your results improve.


I did that too and posted the results in the above post, along with yours results (yours and huhn).

The results with HT off are worse than HT on for me and huhn.


looks like using 3 times the number of threads the CPU got helps for HEVC decoding.

12 thread with core 4 cpu and 12 threads with 4 core cpu with HT threads have about the same speed the min FPS is a lot higher with HT.
at 16 threads it looks like HT give a decent speed boost where non HT has no real difference except min fps.
i guess HT does a good job with a lot of threads too but this will take "a lot" of RAM. i guess even with 32 threads this shouldn't be a real problem for people with 8 thread CPUs.

For me, using HT off and Threads=16 had the same or a little worse results than using Threads=12.

But with HT on, the results of Threads=16 are the best.

Look at the post above with all the results (cyber, me, you)

[NikosD]

Playing around with the number of threads of LAV video I had the chance to answer another question I had regarding hyperthreading and explains the results of cyber, me and huhn.

I finally got the answer I was looking for!

If you have a non hyperthreaded CPU, but you have the ability to change the number of threads of a real parallel process like video decoding, what would be the difference/ benefit of the exact same CPU, with all other parameters equal but Hyperthreading ON ?

The answer came by testing my Core i7-4790 and the other 2160p videos than the ones tested above (beauty and astra)

So, for the 2160p clips 2. and 3. the advantage is 11% - 18%.
It's a clear advantage, but not huge.

It was the transformers clip with the huge bandwidth and a CPU utilization of nearly 100% for (HT on) and (HT off) that gave an advantage to the HT CPU of 27% using 12 threads for both of them.

And this explains how a HT 3.8GHz CPU has the same performance with a non HT 4.4GHz CPU on the transformers clip, using - the HT CPU - a slower OS (Win 8.1) for that task.

On the other hand, for more "serialized" video decoding processes or other similar processes, using a higher clock and a large number of threads, is faster than using a HT CPU with a lower clock.

[NikosD]

i5-3570K @4.4Ghz, 16GB DDR3 1866Mhz 8-9-9-24

UHD_ENT_Transformer_Quad@24fps-51Mbps (10bit)
http://demo-uhd3d.com/files/uhd/Transformers_extinction_Trailer_4K.zip

LAV x64 CPU (Null-P010) 45/59/102
LAV x64 CPU (1280x720 scale-NV12) 45/59/102

LAV x86 CPU (Null-P010) 25/39/86


I used latest nightly MPC-HC x64 v1.7.6.253 using SW decoding with EVR and EVR-CP renderes and while it gives me 0frames dropped and a CPU utilization of 20% - 35%, I see clearly occasional stuttering in movement.

Using your overclocked CPU, are you able to decode this clip in normal/realtime playback using MPC-HC in SW mode, without any "stuttering" or any "glitch" in the movement during the decoding of the video ?

I think it should be the encoding/ muxing of the video and probably the frame-rate which looks like variable although it shouldn't.
It's supposed to be a 23.976fps clips.

I can't see how could this be a decoding performance issue.

Update:

I used DXVA Checker as a player (in play mode) using LAV x64 CPU 0.62.57 and I got:

Frame rate: 18/22/24
CPU utilization %: 18/28/38

[cyberbeing]

You're correct, that sample does seem to have a massive stutter every 2 seconds during playback.

[NikosD]

I think all of them, from that site I posted in a previous post.

I downloaded 6 of them and all of them had the same problem, so I stopped downloading more.

[cyberbeing]

It seems like it could be a problem with FFMPEG, since interestingly enough the NVIDIA Hybrid CUVID decoder (older LAV 2014-09-21 build where 10bit fallback was broken) does not stutter at all even though sections making heavy use of the 10bit color range are very corrupt. Since it would seem those samples are supposedly Samsung showroom demos used to show off the capabilities for the HEVC hardware decoders on Samsung UHD TVs, it would seem rather doubtful that they wouldn't catch playback issues such as this.

[nevcairiel]

It seems like there is a timestamp gap in there whenever that happens. I don't know if its maybe eating a frame or two for some reason, but whenever it stutters the timestamp jumps for 3 frames (125ms) instead of the usual 41.7ms per frame.
Can't tell for sure yet if those frames are missing, or they are not being output for $reasons.

Edit:
Seems like its eating the frames inside the decoder somehow.

[NikosD]

OK.

I've just seen Nevcairiel's answer.

Is it a problem of both x86 and x64 versions ?

[cyberbeing]

Yes, the issue happens on x86 as well.

[NikosD]

It seems like it could be a problem with FFMPEG, since interestingly enough the NVIDIA Hybrid CUVID decoder (older LAV 2014-09-21 build where 10bit fallback was broken) does not stutter at all even though sections making heavy use of the 10bit color range are very corrupt.


Using latest PotPlayer x64 and HEVC DXVA2 decoder for Intel iGPU, I can clearly see the exact same issue like SW decoding.

PotPlayer uses DXVA HEVC for all cases, even 10 bit clips, if you select it.

So, once again even in HW mode, all of my downloaded samples are stuttering.

[nevcairiel]

There is a huge difference between CUVID HW decoding and DXVA decoding, DXVA also uses a lot of ffmpeg code (at least in LAV and since PotPlayer just "borrowed" that code, also in PotPlayer), and CUVID uses none.

[cyberbeing]

The NVIDIA CUVID decoder definitely does not show the same stuttering as the FFMPEG software decoder. The CUVID decoder unlike DXVA is a black box though, so far all we know NVIDIA has some extra error handling logic to avoid "eating the frames inside the decoder" like Nev said is occurring with FFMPEG.

All of those Samsung Demo samples were Main10@L5.2 encoded by ATEME Titan KFE (http://ateme.com/en/products/Transcoders/KFE), and ATEME has a reputation for creating high quality encoding solutions. At this point, I'd give them the benefit of the doubt that these streams are somehow valid, but really who knows. I wouldn't worry too much about it as these are still very early days for HEVC. I'm sure quirks like these will get sorted out by all vendors before 4K Blu-ray is released in Q4 2015.

[nevcairiel]

Its certainly possible the decoder has a bug. I should probably cut a smaller sample out of the Transformers trailer that demonstrates it and poke one of the developers of the decoder with it.
On the other hand, from debug logs it almost seemed like the POC of one frame was repeated, which would indicate the streams have been stitched badly, but i'm not sure i was reading this right.

[cyberbeing]

it almost seemed like the POC of one frame was repeated

Depending on what you are seeing:
All pictures in the same CVS must have a unique POC value. Pictures from different CVSs may share the same POC value, but pictures can still be uniquely identified since there is no possibilities to mix pictures from one CVS with any picture of another CVS. Gaps in POC values are allowed in a CVS...
http://books.google.com/books?id=3WhYBAAAQBAJ&pg=PA35&dq=hevc+POC&hl=en&sa=X

Though it does appear that ATEME Titan KFE is a distributed encoding solution, so bad stitching could very well be within the realm of possibility.

[NikosD]

Its certainly possible the decoder has a bug. I should probably cut a smaller sample out of the Transformers trailer that demonstrates it and poke one of the developers of the decoder with it.



That would be the best.

[P.J]

MPC-HC 1.7.7 LAV 0.63.0, Win 8.1 x64, 750Ti:
"Beauty" video stutters a lot with DXVA2 (EVR/EVR Custom) but no problem with CPU decoder (i3 4150)
"Fitness" works quite great with EVR Custom :confused:

Edit: Almost no problem with the DXVAChecker's player

[NikosD]

I don't have a working Nvidia card to check it out, but I think 750 Ti is a rather powerful card.
Are you sure that DXVA native is selected and not copy-back ?

Also EVR is lighter than EVR-CP.

[P.J]

Yes, it's mostly between 40-70% without max boost.
Even by using CUVID, there's no problem but it's about twice slower than DXVA2, not suitable for 60fps.
10-bit videos doesn't work on Nvidia cards too.

[P.J]

Ok, switched to the iGPU HD4400 and no more stuttering issue with DXVA2.
But it's quite weak for decoding HEVC, played "Fitness" well ~%90 but failed to play "Beauty" smoothly and crashed on 60fps :(

[cyberbeing]

MPC-HC 1.7.7 LAV 0.63.0, Win 8.1 x64, 750Ti:
"Beauty" video stutters a lot with DXVA2 (EVR/EVR Custom)

I can reproduce this with DXVA2-Native. Using EVR-Custom it's quite noticeable, but not as much with plain EVR. The decoder output appears to stall briefly every second or so. madVR seems to avoid this problem by absorbing these brief delays in its queues. DXVA2 Copy-back & CUVID appear unaffected with all renderers.

[NikosD]

I have tried a lot with an Nvidia optimus laptop to make the DXVA decoder work, without success, so I don't have a clear picture what is really going on with Nvidia.

From the posts of cyberbeing above, I think it was clear that DXVA native decoder is very fast for 770 card and he didn't mention any stuttering problems.

But it seems that Nvidia DXVA native decoder has stuttering problems with EVR/EVR CP, so the benchmark speed is useless.

For DXVA copy-back and CUVID, there are no stuttering issues, but the performance is low especially on non super fast cards.

The problem with DXVA native and Nvidia decoder could be a driver's problem or LAV Video problem.

Based on Nevcairiel's posts, the code of DXVA decoder is the same for Intel and Nvidia.

So if Intel's DXVA decoder works in DXVA, which I can assure, then obviously is Nvidia's driver problem.

So, the only good thing left for Nvidia is the ability of CUVID (and only CUVID) decoder to play those pesky 10 bit clips in SW without stuttering.

It seems that is the only decoder capable of doing that.

[JohnLai]

Speak of Intel and Nvidia DXVA, how about AMD solution other than the pretty much pointless Strongene CPU optimised decoder?

Last I read, when AMD Kaveri was released in January 2014, there was a news about AMD collaboration with Telestream to develop HEVC codec that utilised HSA.
It has been 9 months and nothing is heard about the special HSA codec.

On other news, Intel just release new gpu driver for HD2500/4000 version 15.33.29.3945 dated 9/26/2014. Anyone with these IGPU can check if Intel exposes the HEVC DXVA?

[cyberbeing]

I can reproduce this with DXVA2-Native. Using EVR-Custom it's quite noticeable, but not as much with plain EVR. The decoder output appears to stall briefly every second or so. madVR seems to avoid this problem by absorbing these brief delays in its queues. DXVA2 Copy-back & CUVID appear unaffected with all renderers.

Actually it appears that only CUVID has completely stable playback on the Beauty sample, looking at the EVR-Custom CTRL+J stats.

Software = 0ms jitter (-0.8ms <-> 1.2ms)
CUVID = 0ms jitter (-0.8ms <-> 1.2ms)
DXVA2-Copyback = 4ms jitter (-10ms <-> +20ms spikes)
DXVA2-Native = 8ms jitter (-25ms <-> +35ms spikes)

A similar anomaly affecting DXVA2 can also be seen on the Elecard sample:
http://www.elecard.com/assets/files/other/clips/140803_4k_hm130_4s_sao_dbf_qp27.265

Notably, that Elecard sample has the same mulit-threading scaling issue with software decoding as the Beauty sample, so I wouldn't be surprised the issue with DXVA2 has a similar underlying cause. Though seeing CUVID once again being completely immune is curious. It's hard to make an overall conclusion about this without a larger test set of high bitrate 8bit UHD HEVC samples though, since the issue only seems to affect a subset of HEVC encoders.

[NikosD]

Notably, that Elecard sample has the same mulit-threading scaling issue with software decoding as the Beauty sample, so I wouldn't be surprised the issue with DXVA2 has a similar underlying cause. Though seeing CUVID once again being completely immune is curious. It's hard to make an overall conclusion about this without a larger test set of high bitrate 8bit UHD HEVC samples though, since the issue only seems to affect a subset of HEVC encoders.

Have you tried with Nvidia (DXVA/ CUVID) this clip ?

http://demo-uhd3d.com/files/uhd/Hispasat_UHD_NAB-2014_HEVC.zip

Intel's DXVA crashes in both copy-back and native (or gives black screen with audio only)

[NikosD]

Speak of Intel and Nvidia DXVA, how about AMD solution other than the pretty much pointless Strongene CPU optimised decoder?

Last I read, when AMD Kaveri was released in January 2014, there was a news about AMD collaboration with Telestream to develop HEVC codec that utilised HSA.
It has been 9 months and nothing is heard about the special HSA codec.

Don't have high expectations from AMD.

Everything is slow there.

It took them 4 years to implement 4K H. 264 HW decoding.

They announced it on Q2 2010 and they implemented it on Q3 2014.

[dukey]

It took them 4 years to implement 4K H. 264 HW decoding.
Because there is so much 4k hd material ?

[NikosD]

Basically because they said so.

They said that AMD cards support HW acceleration of 4K H.264 and they lied for 4 years and a half.

[P.J]

Copy-back/CUVID like slow motion for me on 60fps video.
Copy-back causes more problems too, stops for some seconds.
10-bit doesn't work with CUVID.
Do Ivybridge/Geforce 600 offer hybrid HEVC too?

[NikosD]

Copy-back/CUVID like slow motion for me on 60fps video.
Copy-back causes more problems too, stops for some seconds.


That reminds me the issues I had with Nvidia 740M and Optimus.


10-bit doesn't work with CUVID.


What do you mean by that ?
It should fall back to SW, I think.


Do Ivybridge/Geforce 600 offer hybrid HEVC too?

I don't have such systems.

[P.J]

It should fall back to SW, I think.

Yes, I meant that :)

Are all of the HEVC videos 10-bit here? http://demo-uhd3d.com/

[NikosD]

No, I have downloaded 8 bit HEVC clips from there, like the one I posted above which crashes Intel's DXVA.

It's here:
http://demo-uhd3d.com/files/uhd/Hispasat_UHD_NAB-2014_HEVC.zip

If you manage to download it, try Nvidia's DXVA/CUVID too.

It should crash I think

[cyberbeing]

Seems http://demo-uhd3d.com finally came back up after being down this past week. All the previous links in the /uhd/ directory have now been moved to a /4k/ directory:
http://demo-uhd3d.com/files/4k/Hispasat_UHD_NAB-2014_HEVC.zip

[NikosD]

Thanks I didn't check it, I've just posted the original source that I had downloaded it.

If someone manages to download it and try to decode it, I would like to know the results of Nvidia.

[cyberbeing]

That Hispasat_UHD_NAB-2014_HEVC sample plays fine with DXVA2-Native, DXVA2-Copyback, and CUVID on my GTX 770.

DXVA2-Native & CUVID have no jitter, stutter, or other issues.
DXVA2-Copyback shows a 5-6ms jitter in EVR-CP, but no noticeable stutter.

[NikosD]

It crashes my Intel DXVA decoder.

That's another difference between the two dxva decoders.

Thanks.

P.S

Assuming of course they didn't change the file!

[cyberbeing]

Assuming of course they didn't change the file!

Well the current hash of Hispasat_UHD_NAB-2014_HEVC.mkv is:
CRC32: 644311D3
MD5: D4985EEC2C985E91DE7D559C560AB51F
SHA-1: 665574705C2588B2E7977F6EAE3524AA576577AC

MediaInfo:
General
Unique ID : 229628849404647287506461971024399774989 (0xACC0E860196F166C85202086ADE7350D)
Complete name : Hispasat_UHD_NAB-2014_HEVC.mkv
Format : Matroska
Format version : Version 4 / Version 2
File size : 2.13 GiB
Duration : 17mn 11s
Overall bit rate : 17.7 Mbps
Encoded date : UTC 2014-05-09 20:05:46
Writing application : mkvmerge v6.9.1 ('Blue Panther') 32bit built on Apr 18 2014 18:15:44
Writing library : libebml v1.3.0 + libmatroska v1.4.1

Video
ID : 1
Format : HEVC
Format/Info : High Efficiency Video Coding
Format profile : Main@L5.0
Codec ID : V_MPEGH/ISO/HEVC
Duration : 17mn 11s
Bit rate : 17.0 Mbps
Width : 3 840 pixels
Height : 2 160 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 24.000 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Bits/(Pixel*Frame) : 0.085
Stream size : 2.04 GiB (96%)
Default : Yes
Forced : No

Audio
ID : 2
Format : E-AC-3
Format/Info : Audio Coding 3
Format settings, Endianness : Big
Codec ID : A_EAC3
Duration : 17mn 11s
Bit rate mode : Constant
Bit rate : 384 Kbps
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 48.0 KHz
Compression mode : Lossy
Stream size : 47.2 MiB (2%)
Default : Yes
Forced : No


It's possible the Intel decoder crashes just because of the bad trim, since even with software decoding the first 20 frames are corrupt likely since it's missing a keyframe at the start.

[NikosD]

Well the current hash of Hispasat_UHD_NAB-2014_HEVC.mkv is:
SHA-1: 665574705C2588B2E7977F6EAE3524AA576577AC


Exactly the same file.


It's possible the Intel decoder crashes just because of the bad trim, since even with software decoding the first 20 frames are corrupt likely since it's missing a keyframe at the start.

That could be an explanation.

It crashes both MPC-HC (LAV video) and PotPlayer (obviously they have similar code)

[nevcairiel]

I've seen that before, some HW decoders crash when they are lacking a reference frame (or too many reference frames), not really easy to do something about that, though.

[JohnLai]

Copy-back/CUVID like slow motion for me on 60fps video.
Copy-back causes more problems too, stops for some seconds.
10-bit doesn't work with CUVID.
Do Ivybridge/Geforce 600 offer hybrid HEVC too?

As for Ivybridge HD2500/4000 with latest driver, I verify that there is no HEVC DXVA support. (Tested it yesterday). There is no HEVC DXVA exposed as shown in DXVAChecker.

Sigh......and here I am hoping for Intel HD4000 HEVC support so that I can use lavfilter Quicksync in future for the acceleration.

Nev, do you have plan to enable HEVC DXVA with Quicksync in future?

[NikosD]

Sigh......and here I am hoping for Intel HD4000 HEVC support so that I can use lavfilter Quicksync in future for the acceleration.



I was the one who was pressing Intel in their own Intel forums to enable HEVC DXVA in their Pentium Haswell line.

I'm not sure if they finally did it, because I switched to Core i7, but last time I checked it out they hadn't done it.

Judging from the results of HD 4600 inside a Core i7, maybe HEVC DXVA is not a good idea to be enabled in such low performance GPUs.

The same reason goes for HD2500 too.

HD4000 could be better, but only up to 720p and a few easy 1080p clips.


Nev, do you have plan to enable HEVC DXVA with Quicksync in future?


The developer of QuickSync decoder is Eric Gur from Intel.

You should ask him if he wants to add QS support of HEVC hybrid decoder, but I think he has already said no.

He will do it when pure HW solution will be available - more than a year from now.

[P.J]

Right, even HD4400/4600 aren't that powerful for hybrid solution specially at 4K@60

EVR-CP causes stutter even while playing 4K H.264

[NikosD]

EVR-CP causes stutter even while playing 4K H.264

I don't think this is right.

Even with Pentium G3420, I never had any problem with 4K H.264 and EVR-CP, with an exception of 4K120fps which need plain EVR.

ALL the other 4K clips have zero stuttering.

If you have a 4K H.264 clip which stutters with EVR-CP, please share it.

[P.J]

You're right, the Jitter value is about 7 for HEVC but 1 for H.264

[hwti]

I have a GTX970 so here are the results of the hybrid decoder with LAV filters 0.63.0 x64.
I used DXVA processing with 1280x720 like the first post, results were a little better with DXVA benchmark.

Beauty GTX970 : sometimes 46/59/65, sometimes 55/80/86 (CPU usage 9-10%) with higher GPU clock
Beauty i7 3930K : 86/104/109 (CPU usage 56%)

Fitness GTX970 : 67/79/106 (CPU usage 13%)
Fitness i7 3930K : 103/150/158 (CPU usage 76%)

Ducks GTX970 : 106/158/182 (CPU usage 11%)
Ducks i7 3930K : 147/167/175 (CPU usage 88%)

So the hybrid decoder doesn't seem to be very fast, but it allows to reduce CPU usage.
DXVA checker doesn't show the GPU usage, but on Beauty sample GPU-Z shows 60-80% usage with GPU@800Mhz Mem@405Mhz when it decodes at 59fps.
I don't know why the clocks are low, perhaps it's optimized for playback, or there is a bottleneck somewhere.

[NikosD]

So the hybrid decoder doesn't seem to be very fast, but it allows to reduce CPU usage.
DXVA checker doesn't show the GPU usage, but on Beauty sample GPU-Z shows 60-80% usage with GPU@800Mhz Mem@405Mhz when it decodes at 59fps.
I don't know why the clocks are low, perhaps it's optimized for playback, or there is a bottleneck somewhere.

I think your DXVA results are fast, it's that your 6 core SandyBridge-E is too fast and makes DXVA results to seem slower.

Of course Nvidia driver looks like it's not optimised for HEVC decoding and it's not stable yet (that's why the clocks go up and down)

But your CPU results are extraordinary.

Which is your CPU clock ?

Update:

After doing some tests with CPU x64 HEVC decoder and DXVA decoding (Null renderer) the performance gap is huge between DXVA Processing (EVR renderer) and DXVA decoding using iGPU (HD 4600)

Beauty:
LAV x64 0.63 (DXVA processing) 53/65/68
LAV x64 0.63 (DXVA decoding) 80/93/96

But using a fast GPU like Nvidia 770 or 970 the performance gap between Null renderer and EVR renderer would be a lot smaller.

So, your results are not extraordinary using DXVA processing and Nvidia 970 (they would be if you were using internal iGPU)

[P.J]

Beauty video isn't interlaced, so why is Null renderer faster?

[NikosD]

Because EVR/ EVR-CP renderers are a lot heavier than Null when using slow GPUs, even in relatively low resolutions like 1280 X 720, which is not low for iGPUs.

I've been writing about this issue of CPU decoders and the degradation of performance due to slow GPUs using even low complexity renderers like EVR/EVR CP the last two years, at least.

GPU is important even in CPU decoding, when you want to display the decoded images.

[huhn]

Because EVR/ EVR-CP renderers are a lot heavier than Null when using slow GPUs, even in relatively low resolutions like 1280 X 720, which is not low for iGPUs.

I've been writing about this issue of CPU decoders and the degradation of performance due to slow GPUs using even low complexity renderers like EVR/EVR CP the last two years, at least.

GPU is important even in CPU decoding, when you want to display the decoded images.

EVR uses the post processing form the GPU driver this can have a huge effect on dxva results but doesn't matter in real world.

even my hd4000 doesn't have problems with madVR and a 60 fps progressive source when setup right.

[NikosD]

That's one of the first things I always disable.

Post processing filters of the driver.

In order to check what we are talking about, you have to benchmark the same video file using "DXVA decoding" and "DXVA processing" options of DXVA checker with your HD 4000.

One issue e.g is scaling, which has a large performance cost for low GPUs.

Using a real renderer even if you disable everything from post processing filtering of the driver's control panel, you can never avoid the performance penalty of scaling a 4K clip to 1280x720p.

[huhn]

That's one of the first things I always disable.

Post processing filters of the driver.

In order to check what we are talking about, you have to benchmark the same video file using "DXVA decoding" and "DXVA processing" options of DXVA checker with your HD 4000.

why should I when real world playback works fine? what to gain from these informations?

[NikosD]

Because that's what we are talking about.

The performance impact of using real world renderer vs null renderer.

You can only play clips that you can actually play, but which is your limit ?

How could you know where is the upper limit of your system ?

If your system can play a clip using DXVA processing with 100fps and using DXVA decoding with 150fps when the real time frame rate of the clip is 24fps, then of course you will not have a real world decoding problem.

But if the clip is 60fps and you can decode it with null renderer (DXVA decoding) in 75fps but with EVR (DXVA processing) in 50fps, then it's more than obvious you will have a problem playing back in real time that 60fps clip using EVR or a more complicated renderer.

For me the main question about this issue is what exactly slows down a low GPU using EVR instead of Null renderer, but doesn't affect so much a high performance GPU which has same results using either null or EVR renderer.

One example I thought of performance penalty is scaling.

[hwti]

I think your DXVA results are fast, it's that your 6 core SandyBridge-E is too fast and makes DXVA results to seem slower.

Of course Nvidia driver looks like it's not optimised for HEVC decoding and it's not stable yet (that's why the clocks go up and down)

But your CPU results are extraordinary.

Which is your CPU clock ?

On Ducks, the CPU is mostly at its turbo frequency, 3.8Ghz.
With other samples, the CPU usage is lower so the clocks change a lot.

The 3 samples don't use an high bitrate, for example the 4K demo on Hotbird is 3840x2160 50fps 30Mbps.
It's 10 bit, so the hybrid decoder doesn't work.
LAV x64 is now very fast on it, but the gap with LAV x32 is huge :
- It plays fine with MPC-HC x64 (50% CPU)
- stuttering with MPC-HC x32 or DVBViewer with LAV x32 (100% CPU)
DXVAChecker seems to have a few problems with it :
- DXVA processing shows 0 fps
- DXVA decoding doesn't show the video, but averages at 85fps for the difficult parts (at least on my 3 minute capture, perhaps there are slower parts during the whole loop)

[NikosD]

The 3 samples don't use an high bitrate, for example the 4K demo on Hotbird is 3840x2160 50fps 30Mbps.
It's 10 bit, so the hybrid decoder doesn't work.
LAV x64 is now very fast on it, but the gap with LAV x32 is huge :
- It plays fine with MPC-HC x64 (50% CPU)
- stuttering with MPC-HC x32 or DVBViewer with LAV x32 (100% CPU)
DXVAChecker seems to have a few problems with it :
- DXVA processing shows 0 fps
- DXVA decoding doesn't show the video, but averages at 85fps for the difficult parts (at least on my 3 minute capture, perhaps there are slower parts during the whole loop)

Can you upload somewhere a sample ?

[hwti]

Can you upload somewhere a sample ?
Here is a short (25s) sample
http://jheberg.net/captcha/hotbird-4k-demo/

[NikosD]

Here is a short (25s) sample
http://jheberg.net/captcha/hotbird-4k-demo/


LAV x64 is now very fast on it, but the gap with LAV x32 is huge :
- It plays fine with MPC-HC x64 (50% CPU)
- stuttering with MPC-HC x32 or DVBViewer with LAV x32 (100% CPU)
DXVAChecker seems to have a few problems with it :
- DXVA processing shows 0 fps
- DXVA decoding doesn't show the video, but averages at 85fps for the difficult parts (at least on my 3 minute capture, perhaps there are slower parts during the whole loop)

why should I when real world playback works fine? what to gain from these informations?


Excellent sample.

@hwti

Because it's a 10bit clip you have to disable from LAV properties the 10bit output format "P010" because Nvidia driver has problems with 10bit clips.

After that, everything works fine with DXVA Checker in both modes "DXVA Decoding" and "DXVA Processing"

@huhn

Look at the results to see what exactly I was saying about Null/ EVR/ EVR-CP renderers.

With signature system and LAV x64 0.63 using CPU (SW decoding) and threads = 16

Benchmarks:
LAV x64 CPU DXVA Decoding (Null) 47/62/66 CPU Usage 98%

LAV x64 CPU DXVA Processing (EVR) 33/49/52 CPU Usage 95%

The best CPU usage I've seen during SW video decoding.

Playback:
In realworld playback using MPC-HC x64 v1.7.7 and EVR-CP the clip has a jitter of ~9-10ms and it can't go more than 44fps-45fps and a few dropped frames.

Using EVR, the jitter=0ms and video decoding is at 50fps - realtime decoding.

[huhn]

but is your file a real world file? I guess not. if a i7 can decode it easily it is simple an extrem case nothing you can buy or normally create.

I don't see HEVC as a real world problem right now.

[NikosD]

When I put the issue of EVR/ EVR-CP renderer being heavier than Null renderer with low GPUs, I put it in general and not HEVC specifically.

Now, about HEVC specifically I have to say that all those samples from ASTRA and HOTBIRD are real world samples of Satellite broadcast video.

What is more real file than that ?

[huhn]

Now, about HEVC specifically I have to say that all those samples from ASTRA and HOTBIRD are real world samples of Satellite broadcast video.

What is more real file than that ?

I don't see a test channel 99.9% of all TV can't display as a real world problem at least not now.
after UHD BD specs and a way to watch these disc on a PC I see this as a real world problem.

[cyberbeing]

Look at the results to see what exactly I was saying about Null/ EVR/ EVR-CP renderers
...
Benchmarks:
LAV x64 CPU DXVA Decoding (Null) 47/62/66 CPU Usage 98%

LAV x64 CPU DXVA Processing (EVR) 33/49/52 CPU Usage 95%
...
Playback:
In realworld playback using MPC-HC x64 v1.7.7 and EVR-CP the clip has a jitter of ~9-10ms and it can't go more than 44fps-45fps and a few dropped frames.

You're EVR-CP perf is usually low, which suggests either a GPU problem or a MPC-HC settings problem. Try 'Right-click->Renderer Settings->Reset->Reset to Default Renderer Settings' and see if things improve.

No issue like that here on my 3570K @4.4Ghz.

LAV x64 CPU DXVA Decoding (Null) 57/68/72 CPU Usage 98%

LAV x64 CPU DXVA Processing (EVR) 57/67/72 CPU Usage 97%

Solid 50.000fps & 0ms jitter with EVR-CP Playback in MPC-HC x64 v1.7.7

[NikosD]

You're EVR-CP perf is usually low, which suggests either a GPU problem or a MPC-HC settings problem. Try 'Right-click->Renderer Settings->Reset->Reset to Default Renderer Settings' and see if things improve.

Solid 50.000fps & 0ms jitter with EVR-CP Playback in MPC-HC x64 v1.7.7

I did the reset to Default Renderer Settings and nothing changed.

I wouldn't call it a GPU problem, it's a performance issue.

The iGPU HD 4600 is simply not fast enough for EVR-CP and such clips which are in the upper limit of performance of CPU decoding (null).

The same processor with a faster GPU, would decode in software using EVR-CP that clip in realtime with no jitter.

[Littleman]

Hi there,
First post since 2008 (yay! had to create a new account...)

I've just bought a lenovo y50 with an nvidia GTX 860m card linked via optimus to an intel HD 4600 igp.

The laptop has a 4k screen, so I had this crazy idea that I would decode 4K files...

I've tried a variety of files and players but it's often a hit or miss so far.
Big Bucks Bunny (http://distribution.bbb3d.renderfarming.net/video/mp4/bbb_sunflower_2160p_60fps_normal.mp4.torrent) in 4k doesn't play smoothly for instance.
Activating the hevc option in players seems to have no effect whatsoever.

I've started digging further into the matter and it seems that all images from the nvidia card go through intel's igp anyway, except when using the hdmi port (no tv, can't try that). What's bugging me so far is that I can't seem to have a solution to force the nvidia chip to decode video and I don't know if it's a harware limitation / optimus limitation or driver limitation...

Any clue to get me started? (all drivers are up to date - intel generic driver, not lenovo and latest nvidia too).
I'm willing to experiment "exotic" solutions if needed.

[NikosD]

I think you can get more help there, after you read that thread:

http://forum.doom9.org/showthread.php?p=1687796

[Littleman]

I think you can get more help there, after you read that thread:

http://forum.doom9.org/showthread.php?p=1687796

Yeah, thanks for the link. I had already read the thread but found no solutions that work for me there, but I'll post in their thread and see if they can help.

[P.J]

I think the new Nvidia 344.48 driver works better.

[NikosD]

I have installed Windows 10 Technical Preview x64 on an old hardware (laptop) and today I installed latest build 9860.

Win 10 has built-in support of MKV and HEVC decoding but during my tests I found out these:


1) There is no DS support, only MFT, so you can forget all known codecs/players besides Windows Media Player and Microsoft's MFT decoders of course.


2) MKV is working for H.264 codec, but not for H.265. I couldn't manage to decode H.265 .MKV files with Microsoft's MFT decoders.


3) Microsoft's MFT H.265 decoder has only CPU support, no DXVA for HEVC yet.


Also I did some benchmarking tests with my laptop's CPU and DXVA Checker x64 v3.2.1

I used LAV x64 0.63.5 and Microsoft's MFT H.265 decoder using Merom T7600 CPU (Core 2 Duo@2.33GHz) and the Gravity 2048x864 HEVC trailer which is in .mp4 format.


DXVA decoding didn't work with Microsoft's H265 decoder so I used DXVA processing.


Gravity 2048x864 HEVC - DXVA Processing - 1280 x720


Microsoft's MFT H.265 20/34/59 CPU Usage: 90%


LAV x64 0.63.5 15/21/23 CPU Usage: 67%



Well, the numbers speak for themselves!

I can't believe such difference with LAV x64 which is very fast, but Microsoft made a miracle with its own decoder!

P.S
I managed to find another .mp4 clip and the difference is much smaller, but still MS MFT is faster again.

The huge difference of the file above is probably due to the encoding process and the not optimised decoding process of LAV for that specific clip.

[huhn]

Microsoft pays there programmer so this decoder is most likely a "finished" product and updates are more for bug not performance in the future. free decoder like openHEVC will catch up in the feature.
and the decoder in lavfilter for HEVC isn't very fast it is just a lot faster than the older version or a different decoder.

[NikosD]

Win 10 is in very early stages and it would be a very big surprise if HEVC decoder is at final stage.

I'll check again both decoders (LAV, MS MFT) in the future to see their progress on the same and other clips.

[kasper93]

Win 10 is in very early stages and it would be a very big surprise if HEVC decoder is at final stage.

If it is deployed that mean it is finished, and only fixes and smaller changes will be incorporated. It is not like open source, company like Microsoft can't afford spending time on improving something that works. It works, it is fast enough, job done.

[huhn]

they just release 9879. feel free to try it out.

[NikosD]

I have to repeat myself that Win 10 is in early stage of development.

Probably H.265 codec is in early stage too.

It's not in a ready state status.

Only RTM version is final product and still changes happen, even in performance not only bugs, via updates.

One thing that makes me sure that it's not final product, it's the fact that no DXVA acceleration is included in this version.

[NikosD]

Does anyone have .mp4 H265 files, but below 3840x2160 resolution ?

I have only two!

[huhn]

a software decoder has usually nothing to do with the DXVA version. the software for a DXVA decoder is just a software that talks to the "GPU driver".

about mp4 maybe ffmpeg can remux to it.

[huhn]

i did a test with the UHD beauty sample. using native resolution
ms mf 9879 85/94/97 76%
lavfilter x64 63.0 71/87/95 69%

doesn't look that special...

[NikosD]

Check Gravity trailer and you will confirm the performance difference above.

Unfortunately on my old laptop I can't check any 4K file, not even H.264

[huhn]

i don't have that file.

[nevcairiel]

There are plenty optmizations for SSE4 and AVX in ffmpeg/LAV, which a Core 2 Duo wouldn't be able to utilize, so its quite possible the performance gap may shrink quite a bit on a newer CPU.

[NikosD]

I don't think so.

I'm pretty sure that such optimizations are already included in MS MFT.

The performance gap will probably be the same regardless of CPU architecture.

The difference is in the clips and the optimizations of specific HEVC decoding parts of the algorithm.

[huhn]

TEARS OF STEEL remux 720p source http://www.h265files.com/

lav 63.0 with 16 threads 821 fps 66 % CPU usage
lav 63.0 with auto 842 fps 60 % cpu usage
ms mf 732 fps 67 % cpu usage

ShakeNDry_3840x2160_120fps_420_8bit_HEVC_RAW.hevc 300 frames remux to mp4 i got this file on a page with tons of samples in all size and codec even lossless YUV. is a long time ago can't find it right now.

lav auto threads 87 fps 76 % CPU
MS MF 90 FPS 85 % CPU

120 FPS game recoding created by HUHN

lav 63.0 366 FPS 53% CPU
ms MF 363 84%

i don't see a clear winner but huge potential in thread optimization for the decoder in Lavfilter

[NikosD]

I upgraded Win 10 to the latest build 9879 and I got a little performance boost for HEVC decoding:

Microsoft's MFT H.265 20/36/60 CPU Usage: 90%

LAV x64 0.63.5 15/21/23 CPU Usage: 67%

I tried LAV with Auto Threads and 8 threads with no performance difference.

I uploaded the clip tested here, in order more people to check it out.

https://www.sendspace.com/file/1aaex4

[huhn]

lav 63.0 117/159/405 61%
ms mf 150/204/292 72%

something about this file i haven't see something like this with every other file.

[NikosD]

Maybe because it's a 10 bit file.

If you have another one, you could try.

[huhn]

i don't have one time to create one.

[xooyoozoo]

This site (http://demo-uhd3d.com/categorie.php?tag=hevc) has a couple 20+mbps 10-bit 4K HEVC demos: Eutelsat (http://demo-uhd3d.com/files/4k/Eutelsat.Demo.HEVC.ts) and Astra (http://demo-uhd3d.com/files/4k/Astra.SES.Demo.HEVC.ts).

You may need to remux.

[P.J]

Interesting, would you try 10-bit 4K with both MS and LAV?

[huhn]

i try it later. but i haven't found a way to get hevc stream out of the TS. so i may go and simply create a 10 bit video should be faster.

[P.J]

Doesn't WMP play TS HEVC?

[Kurtnoise]

i try it later. but i haven't found a way to get hevc stream out of the TS.
did you try with the last version of TsMuxer ?

[huhn]

did you try with the last version of TsMuxer ?

yes it find the stream with Eutelsat.Demo.HEVC.ts but it can't find the fps of the file and an option for 50 is not part of the GUI.
i tried the CLI didn't got it working.
if i remux the stream to m2ts, mmg detects the video stream as AC3/EAC3 X-).

Astra.SES.Demo.HEVC.ts doesn't work at all the stream is not found only audio.

[huhn]

Doesn't WMP play TS HEVC?

no and would be pointless in this case. i don't need it to work in a player i need it to work with a benchmark tool and that's not the case.

[xooyoozoo]

It played in a Win10 VM through the Video app after

ffmpeg (http://ffmpeg.zeranoe.com/builds/) -i $FILE -map 0:0 -c:v copy -movflags faststart $FILE.mp4

.

Edit: Well, I uploaded a remuxed version of the Astra clip (https://mega.co.nz/#!RdlCCL7Z!MtXILmyRrMHE2n5TZJF6mqV24OWUCY8sYXBDJrqokGo).

[huhn]

astra mp4 remux
ms mf 52/58/66 63%
lav 63.0 57/74/117 77%
that's that i guess.
no i didn't mixed them up

[xooyoozoo]

The results for Astra looks great! But it brings up the question of why LAV does poorer on the 10bit Gravity clip.

something about this file i haven't see something like this with every other file.

So I went through and looked at the parameter sets in a few of the clips here, and compared to the others, that clip (Gravity Trailer) has two "unique" features: weighted-prediction and wavefront parallelism.

I hope that either one isn't the sticking point for the LAV decoder, but just to check, can you also try this clip (https://mega.co.nz/#!dEkTDKIC!BJuMB3Gu-Q_pgMnQ9zR6qrFHsoJFM2HbmCDilGqKblo)? It's a 90-seconds 10bit 16Mbps 4K encode with WPP and weighted-pred.

[huhn]

lav 63.0 46/56/68 93%
ms mf 36/48/54 85%

i guess it's wavefront parallelism then.

[xooyoozoo]

WPP is shorthand for wavefront parallelism, so that feature is already turned on.

In any case, the two 10bit-4K clips's results are enough to allay my doubts about LAV's decoder optimizations.

[huhn]

for some reason lav is a lot faster with UHD content but why.

maybe the cabac is faster in ms mf?

[NikosD]

I tried every clip posted here, but unfortunately as I have said before, my laptop probably due to graphics card or DXVA Checker's bug can't benchmark 4K HEVC files (not even 4K H.264), at least under Win 10.

But I uploaded for you another non 4K HEVC file to run a benchmark.

Once again for me, MS MFT 9879 is faster than LAV

TOS sample

MS MFT x64 9879 11/23/36 CPU: 91%
LAV x64 0.63.5 12/19/31 CPU: 91%

Same CPU utilization, MS MFT faster than LAV (about 21%).

You can download the clip here:
https://www.sendspace.com/file/uq8osh

[NikosD]

So I went through and looked at the parameter sets in a few of the clips here, and compared to the others, that clip (Gravity Trailer) has two "unique" features: weighted-prediction and wavefront parallelism.


How did you do that ?

How did you find the encoding parameters of the Gravity trailer and the other clips ?

Is there a tool ?

Thanks!

[huhn]

lav 63 85/120/136 76%
ms mf 102/131/174 93%

ms mf is faster with this file here but usage a lot more CPU.
this file use a lot of cabac with is avg 31 mbit up to 94 mbit unlike the gravity file that has 5 mbit in it name.

[xooyoozoo]

How did you find the encoding parameters of the Gravity trailer and the other clips ?

Is there a tool ?

You'd need a bitstream analyzer that can read the various VPS/SPS/PPS flags. Pretty much all of them can do that, but you might have to read between the lines to match a flag to a colloquially-termed feature.

[NikosD]

Can you name a few ?

Are there any free analysers capable of doing that ?

[NikosD]

I have a new beta of DXVA Checker and for the first time I can benchmark 4K files with my old laptop under Win 10, in "Decode performance" as it's the new name of DXVA decoding.

I can't do the same thing with DXVA processing or "Playback performance" as it is called now, because my iGPU 945G can only output 4K files using VMR7 and not VMR9/EVR needed by DXVA Checker.

Beauty-2160p@30fps-12.3Mbps

MS MFT 9879 8/15/20 CPU usage: 92%

LAV x64 0.63.5 3/9/11 CPU usage: 95%


Once again MS MFT is faster than LAV (about 67% !) and it seems we have a clear winner, at least for Core 2 Duo HEVC decoding.

[Dethis]

It would be interesting to include samples fron Samsung NX1 in the benchmarks HQ (40Mbps) and ProQuality (80mbps)
http://www.imaging-resource.com/news/2014/11/19/samsung-nx1-4k-video-samples-and-new-full-res-frame-grabs-available-for-do

[NikosD]

I downloaded all 5 clips.

It's a little pity for the Mac users to have a new iMac with a 4GHz quad-core Core i7 and AMD Radeon R9 M295X, but still can't play the clips, according to the reviewer of the article.

With signature system and HD 4600@1500MHz I had problems using DXVA Checker and MPC-HC x64 v1.7.7.128 using HW decoding with Intel's HEVC decoder.

I got black screen or nothing.

But using latest official PotPlayer x64, I got DXVA acceleration for all 5 clips and I could play relatively easy all 5 clips, besides the ProQuality@80Mbps clip.

But even the other 80Mbps clip, could be decoded properly by HD 4600 in HW acceleration and ~20% CPU usage@4.0GHz.

I saw the GPU memory usage to 1.5 GB which is unexplainable because from BIOS I can assign up to 1 GB GPU memory!

Using CPU (SW decoder) all clips can be handled easily by all players.

Here is a benchmark for the most difficult clip with DXVA Checker x64 v.3.2.1 and LAV 0.63.5, using DXVA processing at 1280x720:

yvid_SAM_0230-ProQuality-UHD@30fps-80Mbps


LAV x64 0.63.5 CPU 20/43/47 CPU usage: 86 %

[foxyshadis]

Can you name a few ?

Are there any free analysers capable of doing that ?

HEVC Visa (http://hevcvisa.codecian.com/) is good when it works, but lately that hasn't often been the case. The free trial of CodecVisa long since ran out for me.

I thought there was a way to use gstreamer or ffmpeg to display all data, but nothing is working. I guess there's still room for someone to improve ffprobe.

[NikosD]

Thanks for the reply.

From all the apps you mentioned, I have only tried ffprobe which definitely is not enough for analysing HEVC.

Probably I'll have to wait for a more complete app to come up.

[P.J]

I downloaded all 5 clips.

It's a little pity for the Mac users to have a new iMac with a 4GHz quad-core Core i7 and AMD Radeon R9 M295X, but still can't play the clips, according to the reviewer of the article.

With signature system and HD 4600@1500MHz I had problems using DXVA Checker and MPC-HC x64 v1.7.7.128 using HW decoding with Intel's HEVC decoder.

I got black screen or nothing.

But using latest official PotPlayer x64, I got DXVA acceleration for all 5 clips and I could play relatively easy all 5 clips, besides the ProQuality@80Mbps clip.

But even the other 80Mbps clip, could be decoded properly by HD 4600 in HW acceleration and ~20% CPU usage@4.0GHz.

I saw the GPU memory usage to 1.5 GB which is unexplainable because from BIOS I can assign up to 1 GB GPU memory!

Using CPU (SW decoder) all clips can be handled easily by all players.

Here is a benchmark for the most difficult clip with DXVA Checker x64 v.3.2.1 and LAV 0.63.5, using DXVA processing at 1280x720:

yvid_SAM_0230-ProQuality-UHD@30fps-80Mbps


LAV x64 0.63.5 CPU 20/43/47 CPU usage: 86 %
HD4600 can't use more than 512MB as dedicated memory.
And the last clip (ProQuality) doesn't play good with PotPlayer.
(i3 4330) CPU: 25% GPU: 50% Dynamic memory usage: 850MB Dedicated memory usage: 24MB

[NikosD]

No, HD 4600 can use 1GB RAM, that's the max setting of BIOS and if you see my first post with the benchmark details, I have specific numbers for GPU RAM utilisation during benchmarking.

I did the test again monitoring real main memory RAM of my system and GPU RAM and during playing back of one of the samsung clips, GPU RAM went as high as 1.5GB

About Pro quality clip, did you read my post you quoted ?

[xooyoozoo]

Can you name a few ?

Are there any free analysers capable of doing that ?

Intel, Solveig, Parabola all have trials.

I don't think there are any free analyzers for looking at parameter sets. There was a GPL'ed tool a while back that dumped AVC parameter sets, but I don't think the creators have made a HEVC version.

[NikosD]

Thanks for the reply.

I tried all 3 demos of Intel, Solveig, Parabola on Gravity HEVC trailer and none of them worked.

They all have such restrictions in demo apps, that didn't allow them to provide the encoding parameters (or even open the video file in case of Intel)

[P.J]

Ok, Samsung NX1 clips didn't run with 750Ti using PotPlayer.
I also tried 10-bit Samsung UHD clip with 750Ti but it looks like this: :(

http://7.t.imgbox.com/58q8JLof.jpg (http://imgbox.com/58q8JLof)


No, HD 4600 can use 1GB RAM, that's the max setting of BIOS and if you see my first post with the benchmark details, I have specific numbers for GPU RAM utilisation during benchmarking.

I did the test again monitoring real main memory RAM of my system and GPU RAM and during playing back of one of the samsung clips, GPU RAM went as high as 1.5GB

How do you check the GPU RAM?

[ndkamal]

I have an ultrabook with Geforce 840M, and I can't play HEVC files with this chipset, when I try, I have a strange picture like this.

http://9.t.imgbox.com/PRf8wg3f.jpg (http://imgbox.com/PRf8wg3f)

And on the decoder device section, I have not all the formats supported by the Geforce 840M :

http://7.t.imgbox.com/kaqmbOTj.jpg (http://imgbox.com/kaqmbOTj)

Apparenty Optimus has problem with Maxwell chipset. :-(.

[mzso]

Hello!
Is there an OpenCL decoder that does anything useful with an NVIDIA card (gts450)?
I tried the Lentoid HEVC Decoder (It's the same as "Strongene", right? ). The only difference was that it used a shade more CPU.

[wanezhiling]

Strongene PC HEVC/H.265 Decoder Plugin For Intel Media SDK (new) (http://www.strongene.com/en/downloads/downloadCenter.jsp)

[JohnLai]

Great news. Credit to Anandtech article http://www.anandtech.com/show/8811/nvidia-tegra-x1-preview

" For video encode and decode, we see support for 4K60 H.265 and VP9, with support for 10 bit color on decode on H.265"

10 bit as well? I wonder if upcoming GTX960 supports it......

[mzso]

Strongene PC HEVC/H.265 Decoder Plugin For Intel Media SDK (new) (http://www.strongene.com/en/downloads/downloadCenter.jsp)

If you addressed this to me I don't know what you mean.

It's an SDK which is of no use to me.

[nevcairiel]

10 bit as well? I wonder if upcoming GTX960 supports it......

Doubtful, its the same architecture as the 970/980, which don't even have full 8-bit support yet, and it would be nothing but a miracle if they update the hardware decoder in the 960 just 2 month after the earlier launch.

However, the existence of a full HW 10-bit decoder will at least mean it'll be available in the next gen.

[nevcairiel]

Your quote is still speculation, not a fact that the 960 will have it though.

[NikosD]

How do you check the GPU RAM?


I mainly use the gadget of HWiNFO, HWiNFO monitor.
It has same figures as GPU-Z


Hello!
Is there an OpenCL decoder that does anything useful with an NVIDIA card (gts450)?
I tried the Lentoid HEVC Decoder (It's the same as "Strongene", right? ). The only difference was that it used a shade more CPU.

Strongene and Lentoid are the same thing, but Strongene/Lentoid has two kind of decoders.

CPU decoder only and Hybrid CPU/ GPU (OpenCL) decoder.

Also, you could try latest PowerDVD, I think it has an option for OpenCL H.265 GPU assisted video decoding.


Strongene PC HEVC/H.265 Decoder Plugin For Intel Media SDK (new) (http://www.strongene.com/en/downloads/downloadCenter.jsp)


Thanks.

It seems that this is not a stand-alone decoder, but a plugin for developer's use and integration in Intel's MSDK.

You could post it to Eric or Nevcairiel to tell us their opinions about how that could be useful.

[NikosD]

Tegra X1 with its 4K60 fps HW decoder for VP8/VP9 and H.264/H.265 plus 10bit H.265 HW support, is a very interesting product.

I think it's possible for 960 card to have such decoder inside.

We have to wait and see later this month (Jan 22 - I think) for that product arrival.

[mzso]

Strongene and Lentoid are the same thing, but Strongene/Lentoid has two kind of decoders.

CPU decoder only and Hybrid CPU/ GPU (OpenCL) decoder.

Also, you could try latest PowerDVD, I think it has an option for OpenCL H.265 GPU assisted video decoding.


I see. I only noticed in the add filter dialog Lentoid HEVC Decoder and Lentoid HEVC Source (for raw HEVC streams?). So does it have some totally different name? Because only these two start with Lentoid.

Powerdvd's decoder is not available to other apps?

[P.J]

Also, you could try latest PowerDVD, I think it has an option for OpenCL H.265 GPU assisted video decoding.

I don't think...

[ibius]

Cyberlink Video Decoder from PowerDVD v14 does support OpenCL H.265 hw-assisted video decoding and it is available to other DS players, it just has a 'do not use' merit.
You also have to set the decoder to use HAM-mode, don't remember if DXVA-mode would also use OpenCL hybrid decoding. When playing a HEVC video go to the Filters-tab, press and hold 'Ctrl'+'Left Mouse' on the 'Cyberlink Video Decoder (PDVD Generic)', it will open advanced/detailed settings view, the Profile column will tell you if the decoder is using OpenCL or not.
It wont use OpenCL on my GTX 550Ti, but does work on my brothers AMD Radeon HD 5700, too bad, because mine does a lot better on all my samples and tests.
But even in SW-mode I think it does even better than Lentoid and it also supports 10-bit, which Lentoid does not, and both are a LOT faster than LAVVideo x86, on my brothers x64-machine also LAVVideo x64 was slower than Cyberlink and Lentoid, of course it has an old AMD Phenom X4 2.6 Ghz with up to SSE4A instructions.

P.S. last LAV build I tested was 0.63.0.12, don't know how much faster 0.63.0.41 is, I did read it should have better error handling now.

[P.J]

CLCvd.ax? No luck with 750Ti too :(

http://i.imgur.com/h9FyKY0.png

[mzso]

Cyberlink Video Decoder from PowerDVD v14 does support OpenCL H.265 hw-assisted video decoding and it is available to other DS players, it just has a 'do not use' merit.
You also have to set the decoder to use HAM-mode, don't remember if DXVA-mode would also use OpenCL hybrid decoding. When playing a HEVC video go to the Filters-tab, press and hold 'Ctrl'+'Left Mouse' on the 'Cyberlink Video Decoder (PDVD Generic)', it will open advanced/detailed settings view, the Profile column will tell you if the decoder is using OpenCL or not.
It wont use OpenCL on my GTX 550Ti, but does work on my brothers AMD Radeon HD 5700, too bad, because mine does a lot better on all my samples and tests.
But even in SW-mode I think it does even better than Lentoid and it also supports 10-bit, which Lentoid does not, and both are a LOT faster than LAVVideo x86, on my brothers x64-machine also LAVVideo x64 was slower than Cyberlink and Lentoid, of course it has an old AMD Phenom X4 2.6 Ghz with up to SSE4A instructions.

P.S. last LAV build I tested was 0.63.0.12, don't know how much faster 0.63.0.41 is, I did read it should have better error handling now.

I finally figured out that I was using the software decoder. Here the OpenCL decoder just causes an instant crash.

Hmmm. I don't see the cyberlink decoder appearing anywhere in the filters window...

[ibius]

CLCvd.ax? No luck with 750Ti too :(

Yes CLCvd.ax, so it looks like both Cyberlink and Strongene (says so on their site) implemented HEVC OpenCL partial hw-acceleration only for AMD, yet.?
I don't know if its because of NVIDIA's lagging behind OpenCL performance or what, but AMD systems have newer OpenCL version as well.

I finally figured out that I was using the software decoder. Here the OpenCL decoder just causes an instant crash.

Hmmm. I don't see the cyberlink decoder appearing anywhere in the filters window...

Do you mean 'Strongene/Lentoid OpenCL HEVC Decoder'? If so same here, on my brothers AMD machine (HD 5700) Strongene's OpenCL decoder is a mess, out of about two dozen samples it did play 1, I think, without crashing.

Try setting higher merit for 'Cyberlink Video Decoder (PDVD Generic)' globally or just within your player, at least 00600000 or 00800000, and block those that interfere.

[nevcairiel]

Yes CLCvd.ax, so it looks like both Cyberlink and Strongene (says so on their site) implemented HEVC OpenCL partial hw-acceleration only for AMD, yet.?
I don't know if its because of NVIDIA's lagging behind OpenCL performance or what, but AMD systems have newer OpenCL version as well.

NVIDIA and Intel support partial acceleration through DXVA2, which is a far more favorable solution than a proprietary OpenCL decoder, because once they support full acceleration as well, software will not need to change at all.

[NikosD]

Nev, what about that HEVC plugin for Intel's MSDK by Strongene that Wanezhiling posted above ?

Could that be useful and easy to integrate it in LAV filters like Quick sync decoder ?

[nevcairiel]

Its a software decoding module. No interest.

[Yups]

HEVC Hybrid decoding works quite on my Haswell GT2. There are some limitations though. If the GPU load is very high it stutters. This video for example: bbb-3840x2160-cfg02.mkv
http://www.libde265.org/downloads-videos/

It stutters, even though CPU load is relatively low roughly 20%. GPU load is over 90% on this video. 3840x2160 60 fps is too much.

[Yups]

I saw the GPU memory usage to 1.5 GB which is unexplainable because from BIOS I can assign up to 1 GB GPU memory!




It doesn't matter how much of your system RAM is dedicated to your iGPU, Intel GPUs can use up to 1.6-1.7GB of your system RAM.

[NikosD]

This is a new updated post in order to include the CyberLink's PowerDVD v14 H.265 decoder, which is both a 32bit CPU and Hybrid (OpenCL) decoder.

Also, I added a 10bit HEVC clip and Microsoft's MFT Windows 10 HEVC decoder in both x86 and x64 versions.

I used latest versions of DXVA Checker x86/x64 v3.3.1 in benchmark playback performance mode at 1280x720 scaling resolution, LAV x86/x64 0.63.41, and Win 10 beta build 9879

Tests include five systems and the first one has a modern CPU Core i7-4790 with Win 8.1 x64 and Intel HD 4600 iGPU@1.5GHz with v.3960 drivers.
I used Threads=16 for Core i7-4790 using LAV x86/x64 CPU decoder.


Core i7-4790(AVX2) - iGPU HD4600


1.ProRes-1080p@30fps-2Mbps


LAV x64 CPU 177/247/264 CPU 78% Power:55W

LAV x86 CPU 125/151/198 CPU 91% Power:58W

LAV x64 DXVA 100/151/176 CPU 15% Power:50W

LAV x86 DXVA 99/149/168 CPU 17% Power:51W

PowerDVD OpenCL 95/135/152 CPU 23% Power:56W

PowerDVD CPU 79/132/150 CPU 32% Power:36W

Lentoid OpenCL 90/111/120 CPU 20% Power:36W

Lentoid CPU 86/108/110 CPU 16% Power:35W



2.Gravity - 2048x858@24fps-5Mbps-10bit


LAV x64 CPU 105/148/225 CPU 65% Power:51W

LAV x86 CPU 169/106/227 CPU 68% Power:51W

PowerDVD CPU 66/101/197 CPU 55% Power:42W



3.Beauty-2160p@30fps-12.3Mbps


LAV x64 CPU 51/64/66 CPU 80% Power:50W

Lentoid OpenCL 25/43/43 CPU 42% Power:55W

LAV x86 CPU 24/41/46 CPU 93% Power:39W

LAV x64 DXVA 35/37/40 CPU 14% Power:45W

LAV x86 DXVA 34/36/38 CPU 15% Power:45W

Lentoid CPU 10/33/35 CPU 24% Power:40W

PowerDVD CPU 23/29/31 CPU 32% Power:33W


Second system is a Sandybridge Core i5-2400 CPU under Win 8.1 x64 using an ATI Radeon 5750 (14.12 drivers) card with OpenCL support.
I used LAV x86 with Threads=Auto (~1% difference than threads=16), but I used LAV x64 with Threads=16 (more than 20% faster than Auto)



Core i5-2400(AVX) - ATI HD5750


1.ProRes-1080p@30fps-2Mbps


LAV x64 CPU 226/269/318 CPU 91%

Lentoid CPU 128/154/202 CPU 55%

PowerDVD CPU 80/149/194 CPU 52%

LAV x86 CPU 78/108/165 CPU 91%

Lentoid OpenCL 87/107/127 CPU 49% GPU 89%

PowerDVD OpenCL 64/105/141 CPU 23% GPU 83%



2.Gravity - 2048x858@24fps-5Mbps-10bit


LAV x64 CPU 79/138/290 CPU 91%

PowerDVD CPU 56/96/285 CPU 66%

LAV x86 CPU 51/86/264 CPU 89%



3.Beauty-2160p@30fps-12.3Mbps


LAV x64 CPU 43/62/75 CPU 88%

Lentoid CPU 23/38/40 CPU 61%

PowerDVD CPU 25/33/36 CPU 48%

LAV x86 CPU 20/29/31 CPU 91%

Lentoid OpenCL 8/27/30 CPU 47% GPU 80%

PowerDVD OpenCL 18/26/33 CPU 40% GPU 52%



Third system is an old but quad core CPU Core 2 Quad@2.26GHz (266MHzx8.5) with no OpenCL option under Windows 8.1 x64 system using a Fermi Nvidia card - GT610 (347.09 drivers)
I tested a 4K clip, although it's a rather slow CPU not really capable of even 4K@24 fps with any of the decoders.
I used Threads=12 for Core 2 Quad using LAV x86/x64 CPU decoder.

The pure decode benchmarks results, mainly due to slow Nvidia GT610 card and PCI-E v1.0 x4 motherboard, are ~65% faster than the below playback benchmark mode results.
So, for quad cores with better m/b and GPU than mine, the results will be a lot faster in playback mode than these.


C2Q@2.26GHz(SSE4.1) - GT610



1.ProRes-1080p@30fps-2Mbps


LAV x64 46/53/54 CPU 59%

Lentoid 26/45/46 CPU 49%

LAV x86 35/43/52 CPU 90%

PowerDVD 34/38/38 CPU 50%



2.Gravity - 2048x858@24fps-5Mbps-10bit


LAV x64 25/40/58 CPU 85%

LAV x86 23/35/58 CPU 87%

PowerDVD 22/31/38 CPU 69%



3.Beauty-2160p@30fps-12.3Mbps


LAV x64 4/13/13 CPU 62%

LAV x86 CPU 2/12/13 CPU 91%

Lentoid 1/11/12 CPU 49%

PowerDVD 3/9/9 CPU 53%


Fourth system is an old CPU Core 2 Duo - E7300 CPU@2.83GHz (283MHz x 10) with no OpenCL option under Windows 8.1 x64 system using a Fermi Nvidia card - GT440 (347.09 drivers)
The pure decode benchmarks results, mainly due to PCI-E v1.0 x4 motherboard, are a lot faster than the below playback benchmark mode results.
I didn't test any 4K clips, because it's a slow CPU not really capable of even 4K@24 fps with any of the decoders.



C2D@2.83GHz(SSE4.1) - GT440



1.ProRes-1080p@30fps-2Mbps


LAV x64 35/43/50 CPU 86%

Lentoid 29/41/48 CPU 80%

PowerDVD 28/36/39 CPU 85%

LAV x86 20/27/40 CPU CPU 92%



2.Gravity - 2048x858@24fps- Mbps-10bit


LAV x64 20/30/52 CPU 92%

PowerDVD 17/26/41 CPU 83%

LAV x86 14/24/51 CPU CPU 92%


Fifth system is an old CPU Core 2 Duo - T7600 CPU@2.33GHz with no OpenCL option under Windows 10 build 9879 OS.
I didn't test any 4K clips, because it's a slow CPU not really capable of even 4K@10 fps with any of the decoders.


C2D@2.33GHz(SSSE3)-T7600



1.ProRes-1080p@30fps-2Mbps


PowerDVD 22/34/50 CPU 60%

Lentoid 24/31/40 CPU 54%

LAV x64 21/30/54 CPU 89%

LAV x86 19/27/55 CPU CPU 89%



2.Gravity - 2048x858@24fps-5Mbps-10bit


MS MFT x64 21/34/57 CPU 88%

MS MFT x86 19/32/57 CPU 87%

PowerDVD 13/24/56 CPU 70%

LAV x64 16/21/22 CPU 67%

LAV x86 13/20/23 CPU CPU 74%



Comments

There are no Hybrid 10bit HEVC decoders yet, so only CPU can decode 10bit clips.

Intel's Haswell iGPUs (don't know about Ivy) are the only GPUs capable of using both types of hybrid decoders (DXVA and OpenCL)
AMD uses OpenCL only (no DXVA) and Nvidia uses DXVA only (no OpenCL) and only on newer cards.

Lentoid (Strongene) seems to be the fastest x86 decoder, but not with HT CPUs. It looks like a problem with multi-threading. Also is the only decoder not capable of 10bit decoding.

MS MFT Windows 10 decoder can be used only with MP4 container and looks like by far fastest than anything, at least using older SSEx CPUs

PowerDVD v14 decoder can't use hybrid OpenCL decoder for clips above 1080p, at least on my Intel system, but with Radeon 5750 worked fine at that resolution.

LAV x86 has almost always the best CPU utilization (most of the times near 90% on any CPU), but due to different optimizations of LAV x64, the latter is always faster.

On a new modern CPU like Core i7-4790, LAV x64/x86 seems faster than anything else even in both CPU and hybrid decoders. Especially LAV x64 CPU is by far faster than anything (haven't tried MS MFT decoder).

Lentoid OpenCL decoder is very fast on 4K clips.

PowerDVD v14 is the slowest decoder of all on 4K clips, but faster than Lentoid on 1080p.

On the other hand, for an old CPU like Core 2 Duo with SSSE3, MS MFT is clearly ahead of all other decoders regarding CPU performance.
PowerDVD is faster than both LAV and Lentoid.

LAV in both versions is the slowest decoder and can't keep up in realtime decoding for both clips using x86 version on a Core 2 Duo processor. LAV x64 can't keep up for Gravity, too.

The Core i5-2400 system manages to be faster than Core i7-4790 (!) on the easier to decode 1080p clip and very close to 4K clip decoding, due to the faster discrete card (Radeon 5750) than the iGPU of Core i7.

Once again the importance of a fast card is obvious when using CPU decoding real-world benchmark tests, using a renderer in playback performance mode and not pure decoding performance with a null renderer.

Using a 45nm Core 2 Duo with SSE4.1, LAV x64 is once again faster and LAV x86 is the slowest.

For quad core SSE4.1 CPUs, PowerDVD doesn't scale good, so it's once again the slowest of all decoders, where LAV x64 is once more faster, but not far from LAV x86 and Lentoid.

[NikosD]

Its a software decoding module. No interest.

It doesn't matter how much of your system RAM is dedicated to your iGPU, Intel GPUs can use up to 1.6-1.7GB of your system RAM.

Thanks.

[mzso]

Do you mean 'Strongene/Lentoid OpenCL HEVC Decoder'? If so same here, on my brothers AMD machine (HD 5700) Strongene's OpenCL decoder is a mess, out of about two dozen samples it did play 1, I think, without crashing.

Try setting higher merit for 'Cyberlink Video Decoder (PDVD Generic)' globally or just within your player, at least 00600000 or 00800000, and block those that interfere.

Even with AMD? Huh...

It doesn't even appear in the available codecs list neither in Potplayer nor in MPC-HC. (checked with x64 mpc-hc too) I don't know why. I installed pdvd. Do you know which is the codec's file and what are it's dependencies? Maybe I could add it manually....

[ibius]

Even with AMD? Huh...

It doesn't even appear in the available codecs list neither in Potplayer nor in MPC-HC. (checked with x64 mpc-hc too) I don't know why. I installed pdvd. Do you know which is the codec's file and what are it's dependencies? Maybe I could add it manually....

Well at least with older AMD HD5700, but according to NikosD's test results on the first page Strongene's OpenCL runs on iGPU HD4600 as well, although not much better, it crashed on 3 samples out of 6.
Nev says Intel supports partial DXVA2 decoding like NVIDIA, but both Strongene's and Cyberlink's OpenCL implementation works on AMD and Intel but not even on 750Ti, so I guess NVIDIA's OpenCL is for sh*t.

No need to check MPC-HC x64, Cyberlink Video Decoder is 32-bit only like PowerDVD itself.
If you don't even see the filter, find CLCvd.ax and register it manually, with PowerDVD installed you should have the decoder and it's dependencies in PowerDVD\Common\VideoFilter folder.

[mzso]

Well at least with older AMD HD5700, but according to NikosD's test results on the first page Strongene's OpenCL runs on iGPU HD4600 as well, although not much better, it crashed on 3 samples out of 6.
Nev says Intel supports partial DXVA2 decoding like NVIDIA, but both Strongene's and Cyberlink's OpenCL implementation works on AMD and Intel but not even on 750Ti, so I guess NVIDIA's OpenCL is for sh*t.

No need to check MPC-HC x64, Cyberlink Video Decoder is 32-bit only like PowerDVD itself.
If you don't even see the filter, find CLCvd.ax and register it manually, with PowerDVD installed you should have the decoder and it's dependencies in PowerDVD\Common\VideoFilter folder.

Cool. That helped. The HAM setting should be the one using OpenCl, right? Not sure if it's doing it though. It's maybe a shade faster than lav based on CPU usage, hard to tell. And I don't see any GPU usage increase.

[P.J]

so I guess NVIDIA's OpenCL is for sh*t.

Well explained :p :thanks:

[NikosD]

Is there someone who knows what's going on with DivX 10 HEVC decoder ?

Latest version 10.2.4 has removed Codec Pack and after installation, it downloads the necessary HEVC codec on demand.
Furthermore, doesn't expose this DS filter (?) to 3rd party apps.

After DivX 10.2.4 installation, it blocked every x86 codec installed (!) and couldn't load anything in DXVA Checker.

If someone could tell me how could I use DivX 10 HEVC codec, I would definitely test it :)

[Yups]

Interesting for Haswell and Broadwell.

- Improved video playback through partial hardware acceleration support for the VP9 video format.
- GPU accelerated decode of HEVC video file format including both 8-bit and 10-bit support. This will provide improved video playback capabilities on the platforms.
https://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&DwnldID=24596

[NikosD]

I tried that driver and doesn't look to have a DXVA decoder device for VP9.
Unless it's not named as VP9 or maybe DXVA Checker doesn't recognize it.
http://s15.postimg.org/71k4hap0b/DXVA_Checker.png

I also tried 10bit HEVC clips with latest LAV Video 0.63.41 HEVC DXVA decoder and I only get a black screen with no decoding.

[Yups]

What was your previous driver?

[NikosD]

v. 3960.

Why ?

Are you looking for comparisons between drivers ?

[Yups]

Because I noticed some differences from my previous driver (4029).

This video: http://demo-uhd3d.com/files/uhd4k/Demo_Samsung_2014_-_Lovely_Swiss.zip

With driver build 4029 GPU usage was low and CPU usage 80-100%. With driver 4080 I get 40-50% CPU usage and 90-100% GPU usage in Potplayer (colour issues unfortunately), in MPC Home Cinema there is only a black screen.

This video: bbb-3840x2160-cfg02.mkv

GPU usage is higher than before, 99-100% usage now. Previous usage was ~80%.

[wanezhiling]

driver 10.18.14.4080

Just like a slide show...
:o:o
http://i3.tietuku.com/21a83a47c35363e0.png

[JohnLai]

10 bit HEVC dxva decoding by intel? I not sure if intel Haswell HD IGPU has the hardware to decode it in first place.

This one probably refers to upcoming Broadwell.

^.^ If this 10bit dxva for haswell igpu really work, somebody please inform NVIDIA that their competitor already has 10bit dxva working. AMD too.

[nevcairiel]

From what I've seen its still kind of pointless, as it only supports 8-bit output, so while it may be able to decode it partially, it then probably rounds the 10-bit output to 8-bit, which degrades the quality.
I need to check in more detail if its actually possible to get it to output 10-bit.

[NikosD]

10 bit HEVC dxva decoding by intel? I not sure if intel Haswell HD IGPU has the hardware to decode it in first place.

This one probably refers to upcoming Broadwell.

^.^ If this 10bit dxva for haswell igpu really work, somebody please inform NVIDIA that their competitor already has 10bit dxva working. AMD too.

We are not talking about fixed-function HW decoder, like QuickSync, but for partial GPU assisted (via shaders) HW decoding.

So, it's possible to decode 10bit HEVC using a hybrid decoder.

From what I've seen its still kind of pointless, as it only supports 8-bit output, so while it may be able to decode it partially, it then probably rounds the 10-bit output to 8-bit, which degrades the quality.
I need to check in more detail if its actually possible to get it to output 10-bit.

But why the black image and no decoding without fall-back to SW using LAV Video ?

If all modifications are inside drivers, I think it could or should work instantly, without modifications of LAV video.

Or at least fallback to SW mode.

[nevcairiel]

They probably broke something else.

[NikosD]

With driver build 4029 GPU usage was low and CPU usage 80-100%. With driver 4080 I get 40-50% CPU usage and 90-100% GPU usage in Potplayer (colour issues unfortunately), in MPC Home Cinema there is only a black screen.

GPU usage is higher than before, 99-100% usage now. Previous usage was ~80%.

They probably broke something else.

PotPlayer seems to use DXVA for 10bit clips, but with artifacts and colour issues.
But it uses DXVA.

Also, PotPlayer plays fine in DXVA those Samsung NX1 clips posted above http://www.imaging-resource.com/news/2014/11/19/samsung-nx1-4k-video-samples-and-new-full-res-frame-grabs-available-for-do, when LAV DXVA native crashes immediately.

Also, LAV DXVA copy-back doesn't crash, but falls back to SW decoding using the above NX1 clips.

[nevcairiel]

PotPlayer seems to use DXVA for 10bit clips, but with artifacts and colour issues.
But it uses DXVA

I deliberately disabled 10-bit from even attempting to use DXVA2 because broken decoding is worse than no decoding. LAV will never knowingly produce wrong output.

HEVC 10-bit has a special second decoder device called HEVC_VLD_Main10 (or {107AF0E0-EF1A-4D19-ABA8-67A163073D13}). If any GPU supports proper 10-bit decoding, I expect it to offer this device. Its the only way to really know if it supports it or not, without hard-coding a list of GPUs, which I would be very reluctant to do.
In fact, the DXVA2 HEVC specification clearly states the the HEVC_VLD_Main device shall only be used for the "Main" profile, and HEVC_VLD_Main10 for the "Main10" profile.

So far I have not seen this new device on any GPU, even with Intels new driver which presumably claims 10-bit support.

[NikosD]

OK, but Samsung NX1 clips are not 10bit, they are 8bit.

[wizziwig]

So far I have not seen this new device on any GPU, even with Intels new driver which presumably claims 10-bit support.

Only on Broadwell. Haswell is limited to 8-bit only.

https://communities.intel.com/message/273816#273816


10bit hybrid HEVC decode is enabled on BDW U and H sku. Not available on HSW.

8bit hybrid HEVC decode is available on both HSW U sku and above and BDWU and H sku.

Thanks
Robert

[Yups]

Ok that makes sense. No wonder there are playback issues on my HSW with 10 bit. Looks to me that the playback improvements refer to Broadwell.

[nevcairiel]

Thats unfortunate, as I'm unlikely to get a Broadwell-U CPU. Maybe once a desktop Broadwell comes out, or maybe the rumors are true and NVIDIA has 10-bit support in the 960, and working on it there may work on Intel as well.

[NikosD]

Only on Broadwell. Haswell is limited to 8-bit only.

https://communities.intel.com/message/273816#273816

Ok that makes sense. No wonder there are playback issues on my HSW with 10 bit. Looks to me that the playback improvements refer to Broadwell.

Thanks wizziwig.

I really like the speed of posting in Intel forums by Yups [emoji2]

The Broadwell thing crossed my mind, but I didn't want to think that, that was the reason.

Probably VP9 HW acceleration refers to Broadwell only as well.


Thats unfortunate, as I'm unlikely to get a Broadwell-U CPU. Maybe once a desktop Broadwell comes out, or maybe the rumors are true and NVIDIA has 10-bit support in the 960, and working on it there may work on Intel as well.

Nobody with a decent Haswell is going to get a Broadwell-U.

Personally, I'll move directly to Skylake, skipping even desktop Broadwell, even if Skylake is not going to be released along with desktop Broadwell as Intel says.

In my opinion, Skylake will come after desktop Broadwell due to the well known Intel's delays, but that's only a speculation.

[Yups]

Broadwell-K might be interesting to see how GT3e performs but Skylake-S with its Gen9 has a newer Gfx architecture, it should support HEVC decode in hardware finally. This is much more appealing.


In my opinion, Skylake will come after desktop Broadwell due to the well known Intel's delays, but that's only a speculation.


Broadwell-K and Skylake-S are both scheduled for H2 2015. Broadwell-K might come 2-3 months earlier though.

[JohnLai]

Only on Broadwell. Haswell is limited to 8-bit only.

https://communities.intel.com/message/273816#273816

>.< Deep inside my heart, I really hope for Haswell 10bit support.....I guess this is it...I can only wait for either AMD or Nvidia gpu product with fixed hardware acceleration to come out.

BTW, nobody wanna report the mis-exposure of 10bit HEVC DXVA interface for Haswell IGPU at Intel forum?

[nevcairiel]

OK, but Samsung NX1 clips are not 10bit, they are 8bit.

I fixed the Samsung clips, they work fine now.

For 10bit..
Intel broke their drivers, similar to the problem that NVIDIA had quite recently.

For some reason, EVR tries to accept a 10-bit/16-bit input (P016 in this case), but it fails at it, resulting in overall playback failure.

NVIDIA had the same problem a few weeks back, when a driver update caused P010 to be accepted by EVR, but EVR did not handle it properly and failed.

Someone should bring this to Intels attention.
You can work around this problem by disabling P010 and P016 output in LAV if you use EVR.

[NikosD]

I fixed the Samsung clips, they work fine now.


Yes, it works.

But this 8bit HEVC clip, still crashes Intel's DXVA (both native and cb)

http://demo-uhd3d.com/files/uhd4k/Hispasat_UHD_NAB-2014_HEVC.zip

[nevcairiel]

But this 8bit HEVC clip, still crashes Intel's DXVA (both native and cb)

http://demo-uhd3d.com/files/uhd4k/Hispasat_UHD_NAB-2014_HEVC.zip

This one crashes somewhere deep in the driver, those take some time to figure out. Presumably because its missing a reference frame and Intel expects it to be present.

Ideally, Intel would just fail to decode the frame instead of crashing completely, but from past experience, this is a common theme with Intel .. they crash when something is wrong.

[Gravitator]

ERROR 503 - Service Unavailable!
Привет,
I also have a problem (Firefox).
Use a manual download from > http://demo-uhd3d.com/fiche.php?cat=uhd&id=21

[NikosD]

As for Ivybridge HD2500/4000 with latest driver, I verify that there is no HEVC DXVA support. (Tested it yesterday). There is no HEVC DXVA exposed as shown in DXVAChecker.


It seems that Ivybridge still uses 15.33 driver and not 15.36, so I don't expect Ivy to support DXVA HEVC acceleration.

But have you tried Lentoid HEVC OpenCL decoder or Cyberlink's PowerDVD HEVC OpenCL decoder with Ivy to see of there is any HW acceleration from there ?

Because Ivy supports OpenCL.

[ksio89]

I'm not sure if it goes here, be free to move if not.

I would like to ask you guys: is there any hw decoder available for HEVC? I have MPC-HC (x64 flavour) installed along K-Lite Codec Pack Full 10.9.5, and everything is default, so the default video decoder is LAV, and the renderer is EVC CP.

In LAV Video decoder menu, I choose DXVA2 (native) as the HW decoder, and ticked "HEVC" and "UHD (4K)" boxes, both on LAV x86 and x64 (even though I only use x64 version of MPC-HC).

I downloaded a 4K video sample from here:

http://www.imaging-resource.com/news/2014/11/19/samsung-nx1-4k-video-samples-and-new-full-res-frame-grabs-available-for-do,

The file content generated by MediaInfo is attached as a .txt.

I played this sample in MPC-HC, but DXVA was not activated, so the framerate was choppy. I played a Full HD MP4 video file just to test and DXVA worked fine, low CPU usage and higher GPU usage.

These are the specs of my laptop:

Dell Inspiron 15R-5537:

Intel core i7-4500U @ 2,7 GHz
AMD Radeon HD 8870M (128-bit, 2 GiB GDDR5)
6 GB DDR3L @ 1600 MHz
Windows 8.1 Pro x64

Does this HW support 4K DXVA decoding? It's my laptop that's too weak to play 4K video file, or I can't just set up HEVC DXVA decoding properly? Thanks in advance.

[NikosD]

I'm not sure if it goes here, be free to move if not.

I would like to ask you guys: is there any hw decoder available for HEVC? I have MPC-HC (x64 flavour) installed along K-Lite Codec Pack Full 10.9.5, and everything is default, so the default video decoder is LAV, and the renderer is EVC CP.

In LAV Video decoder menu, I choose DXVA2 (native) as the HW decoder, and ticked "HEVC" and "UHD (4K)" boxes, both on LAV x86 and x64 (even though I only use x64 version of MPC-HC).

I downloaded a 4K video sample from here:

http://www.imaging-resource.com/news/2014/11/19/samsung-nx1-4k-video-samples-and-new-full-res-frame-grabs-available-for-do,

The file content generated by MediaInfo is attached as a .txt.

I played this sample in MPC-HC, but DXVA was not activated, so the framerate was choppy. I played a Full HD MP4 video file just to test and DXVA worked fine, low CPU usage and higher GPU usage.

These are the specs of my laptop:

Dell Inspiron 15R-5537:

Intel core i7-4500U @ 2,7 GHz
AMD Radeon HD 8870M (128-bit, 2 GiB GDDR5)
6 GB DDR3L @ 1600 MHz
Windows 8.1 Pro x64

Does this HW support 4K DXVA decoding? It's my laptop that's too weak to play 4K video file, or I can't just set up HEVC DXVA decoding properly? Thanks in advance.

Your CPU has an iGPU HD 4400 inside, which is capable of HW acceleration of HEVC in both ways - OpenCL and DXVA.

Your discrete card Radeon 8870M has only OpenCL HW accelerated support of HEVC.

LAV and MPC-HC have only DXVA HW acceleration support of HEVC, not OpenCL.

If you enable your iGPU to handle video decoding, you 'll see [Playing H/W], which means HW acceleration, in the left bottom of MPC-HC by enabling Status[Control+5] to see the status of playback decoding.

If Radeon 8870M is the primary card for video playback, don't expect HW acceleration of HEVC using LAV, you have to use Cyberlink's PowerDVD HEVC decoder or Lentoid OpenCL HEVC decoder.

The HW acceleration of full HD MP4 file probably occurs because it's not a HEVC file, but H.264 (AVC) which Radeon 8870M can handle in DXVA HW acceleration.

[NikosD]

Intel put a table to clarify the situation of VP9 partial acceleration and HEVC 10bit partial acceleration for Haswell and Broadwell here:

https://communities.intel.com/thread/59216

Also, there is a post in the above link, referring to Nevcairiel's post here, about broken 10bit HEVC driver for HSW.

[xooyoozoo]

Chromium added DXVA VP8/VP9 decode acceleration support over a month ago (https://codereview.chromium.org/719273003). Based on the revision numbers, it looks like Chrome Dev and Canary should already have the acceleration activated.

Just FYI, in case anyone has Broadwell and wants to entertain our curiosity. ;)

[nevcairiel]

Note that its not actually DXVA being used, but directly loading the special Intel decoder DLLs.

[NikosD]

Chromium added DXVA VP8/VP9 decode acceleration support over a month ago (https://codereview.chromium.org/719273003). Based on the revision numbers, it looks like Chrome Dev and Canary should already have the acceleration activated.

Just FYI, in case anyone has Broadwell and wants to entertain our curiosity. ;)

I would like to test it with my HD 4600, because Intel's drivers 15.36.xx family on June 2014 with beta v.3651 driver, introduced for the first time the VP8 MFT decoder even for Haswell.

The link you provided says:


The decoding is for Intel Broadwell and Bay Trail CPUs, which will include an
onboard hardware accelerated decoder.


IIRC, the QuickSync ASIC inside Haswell iGPU doesn't include VP8 as a HW accelerated codec.

But maybe Chrome uses that "Intel Hardware VP8 decoder MFT" that I see in MFT decoders installed and I could get some kind of HW acceleration using iGPU's shaders and VP8 streaming video.

Does latest Chrome v39 under Win 8.1 have this acceleration enabled ?

I threw a VP8 clip on Chrome and only CPU was used during the decoding.

I don't know if I have to enable a switch or something like that for VP8 HW acceleration in Chrome.

Is there any way to "force" youtube.com to send VP8 streams, because nowadays youtube uses VP9 mainly.
Or maybe change some settings in Chrome for youtube.com in order to send VP8 streams.

Note that its not actually DXVA being used, but directly loading the special Intel decoder DLLs.

I think Microsoft hasn't released VP8/VP9 DXVA decoder devices yet, so I wonder what's the meaning of this phrase from the above link:


Adds code to DXVA to try and load the Intel media decode DLLs


Is this even possible to use DXVA code without DXVA decoder device ?

[nevcairiel]

They just integrated this special Intel DLL code into their DXVA code in Chromium, the commit message is slightly confusing there, its not actually using anything from DXVA2 otherwise.

[NikosD]

I see.

What about Chrome and Haswell ?

Is there any chance that VP8 HW acceleration could be working ?

Update:

I have downloaded and tested latest Chromium v42 x86/x64 (21/01/2015) from here http://chromium.woolyss.com/ and it seems that even playing a demanding 4K VP8 file, from my disk not Youtube, it doesn't trigger any kind of HW acceleration.

Only CPU is involved in VP8 decoding with my signature system.

Someone with Broadwell processor, could tell us more about the subject.

[nevcairiel]

From AnandTech GTX960 launch preview:

When it comes to HEVC (H.265) decoding, GM204, GM1xx, and GK1xx GPUs all offer limited HEVC decoding via a hybrid process that offloads some parts to the GPU fixed function units and shaders, and other parts to software entirely. However as of GM206 NVIDIA’s fixed function HEVC decoder has been completed and rolled into this GPU, making this the first discrete GPU with full HEVC decoding support.


A full review is expected in the following days.

[NikosD]

Right GTPVHD.

I took a quick look on over 10 reviews, nobody has the knowledge or will to test its HTPC/Media capabilities.

My questions are for both H.264/H.265:

1) 10bit decoding
2) Speed, especially on 100+ Mbps 4K HEVC clips

and

3) What about VP8/VP9

[nevcairiel]

There is still no DXVA2 spec for VP8 or VP9, and I have see no mention of it in the CUVID decoder headers, so I don't see how it would magically appear.
Maybe in the CUDA SDK 7, should try to get my hands on that RC.

H.264 is not getting 10-bit decoding, since its not part of the base profile (like it is in H.265).

[nevcairiel]

NVIDIA GTX 960 Benchmarks

1.Beauty-2160p@30fps-12.3Mbps

LAV x86 DXVA2 Native, Decode, 125 fps
LAV x86 DXVA2 Native, Playback (scaling to 1080p), 125 fps

LAV x86 DXVA2 CopyBack, Decode, 118 fps

2.Fitness-2160p@30fps-8Mbps

LAV x86 DXVA2 Native, Decode, 148 fps
LAV x86 DXVA2 Native, Playback (scaling to 1080p), 145 fps

LAV x86 DXVA2 CopyBack, Decode, 140 fps

3.Ducks-2160p@50fps-4Mbps

LAV x86 DXVA2 Native, Decode, 153 fps
LAV x86 DXVA2 Native, Playback (scaling to 1080p), 152 fps

LAV x86 DXVA2 CopyBack, Decode, 143 fps

Because none of these samples posed any challenge to the decoder, I included the "Ultra HD Sample #1" from here: http://www.imaging-resource.com/news/2014/11/19/samsung-nx1-4k-video-samples-and-new-full-res-frame-grabs-available-for-do (the first video with the big leaves)
This video has a much higher bitrate as the others, at nearly 80Mbps.

SamsungNX1.Leaves.2160p@30fps-80Mbps

LAV x86 DXVA2 Native, Decode, 118 fps
LAV x86 DXVA2 Native, Playback (scaling to 1080p), 116 fps

LAV x86 DXVA2 CopyBack, Decode, 115 fps

-----

Its important to note that CPU usage in these tests was always below 1% (except CopyBack), indicating that full hardware acceleration is working perfectly.

From these 8-bit movies, I can conclude that the HEVC/H.265 hardware decoder in the GTX 960 is plenty fast for 4K HEVC content.
It seems to cap out around 150-160 fps for 4K, which is slightly below the Hybrid decoder on "simple" samples, however like any decent hardware decoder its largely unimpressed by high bitrates and retains most of its speed, while the hybrid decoder gets much much slower.

I'll work on implementing 10-bit support in LAV Video soon, and we'll see how it fares there (and hopefully EVR accepts 10-bit DXVA2-Native input)

Edit:
Added CopyBack numbers. Even those have greatly improved, and the performance difference between native and CB is very small (on my system, anyway).

[NikosD]

Thanks for the results.

NVIDIA GTX 960 Benchmarks


3.Ducks-2160p@50fps-4Mbps

LAV x86 DXVA2 Native, Decode, 153 fps
LAV x86 DXVA2 Native, Playback (scaling to 1080p), 152 fps

LAV x86 DXVA2 CopyBack, Decode, 143 fps



The low bitrate clips are definitely not in favor of pure HW decoders.

Nvidia GTX 770 DXVA is faster on this clip.

When I tested DXVA on hybrid decoders LAV x64 was a lot faster than LAV x86.

Of course now with the pure decoder, it shouldn't play a significant role, but if I were you, I would definitely check LAV x64 with pure HW decoder.


SamsungNX1.Leaves.2160p@30fps-80Mbps

LAV x86 DXVA2 Native, Decode, 118 fps
LAV x86 DXVA2 Native, Playback (scaling to 1080p), 116 fps


It's definitely shows its muscles here, but I would like to see DXVA performance of a hybrid decoder here.

Can you switch to hybrid decoder using shaders on 970 or there is no way to control it and you have to use the ASIC ?



Its important to note that CPU usage in these tests was always below 1% (except CopyBack), indicating that full hardware acceleration is working perfectly.


Can you check the video engine usage during benchmark tests ?
Not the GPU load but video engine load.

I use gadgets for this (GPU monitor) but I think GPU-Z shows that too.


From these 8-bit movies, I can conclude that the HEVC/H.265 hardware decoder in the GTX 960 is plenty fast for 4K HEVC content.


It looks fast like 4K 120fps decoder, but if the rumors are true about new 4K HEVC Bluray disks, we have to test it on more than 100Mbps clips up to 150Mbps and I think in 10bit and possibly 60 fps.


I'll work on implementing 10-bit support in LAV Video soon, and we'll see how it fares there (and hopefully EVR accepts 10-bit DXVA2-Native input)


Did your code work as it is or you had to change it for 960 ?

[nevcairiel]

It's definitely shows its muscles here, but I would like to see DXVA performance of a hybrid decoder here.

Can you switch to hybrid decoder using shaders on 970 or there is no way to control it and you have to use the ASIC ?


You don't get a choice, if it supports full HW decoding, thats what its going to do. But luckily I also have a GTX 980, which doesn't have the new decoder.

SamsungNX1.Leaves.2160p@30fps-80Mbps using the hybrid decoder on a GTX 980:
LAV x86 DXVA2 Native, Decode, 22 fps
LAV x86 DXVA2 Native, Playback (scaling to 1080p), 17 fps

LAV x64 DXVA2 Native, Decode, 24 fps
LAV x64 DXVA2 Native, Playback (scaling to 1080p), 21 fps

Not to mention that this mode maxes out one CPU core at 100% usage as well.
High bitrates have a steep performance cost in software and hybrid decoders.


Did your code work as it is or you had to change it for 960 ?
No changes were required for 8-bit decoding.

[NikosD]

NVIDIA GTX 960 Benchmarks

I'll work on implementing 10-bit support in LAV Video soon, and we'll see how it fares there (and hopefully EVR accepts 10-bit DXVA2-Native input)


Does DXVA checker show a DXVA device decoder of HEVC_Main10 for 960 ?

[nevcairiel]

Yes it does, with P010 output instead of NV12, so its a proper 10-bit decoder.

[Stereodude]

Does it have HEVC Version 2 (http://en.wikipedia.org/wiki/High_Efficiency_Video_Coding#Profiles) support?

[nevcairiel]

No, and there are unlikely to ever be hardware decoders that support it. Similar to H.264, only the first set of profiles is supported, as those are the ones used for commercial content. Luckily 10-bit is part of Version 1.
The DXVA2 Spec is only for Main and Main10 profiles.

[Stereodude]

On another forum someone reported that UHD Blu-Ray will require Version 2 support, but I haven't been able to verify that.

[nevcairiel]

The final spec hasn't been confirmed yet, but from what I hear, the "required" features for 4K Blu-ray are 2160p, 4:2:0 in 10-bit and up to 60 fps (Version 1), with higher bitdepths being an optional feature for players. I'm not sure how they want to do optional support for something like this though, unless they ship a second disc or a second video track on the same disc. Previously, HD audio was optional, and there always has been a non-HD fallback mechanism in place. So this optional higher bitdepth should still be taken with a grain of salt.

[sneaker_ger]

Unlikely. UHd Blu-Ray is pretty much confirmed 10 bit 4:2:0 and to be on the market at the end of 2015 while >10 bit HEVC is not finalized yet.

[nevcairiel]

Actually HEVC Version 2 has been finalized last year (October I think) and is included in the latest release of the full spec.

[sneaker_ger]

Really? How did I miss that?

[Stereodude]

Really? How did I miss that?
October 29, 2014 according to Wikipedia.
On October 29, 2014, HEVC/H.265 version 2 was approved as an ITU-T standard.

[Stereodude]

NVIDIA GTX 960 Benchmarks
I don't want to drag this thread off topic, but since you seem to be the only poster with a GTX 960 so far... Have you tested it with madVR? I'd love to know which AMD card it's on par with in terms of shaders power.

[NikosD]

NVIDIA GTX 960 Benchmarks


The DXVA native and copy-back results of Nvidia 960 are really close, but in pure "Decode" mode.

What about "Playback" mode scaled to 1080p or 720p ?

Are they still close in that mode ?

[xooyoozoo]

The final spec hasn't been confirmed yet, but from what I hear, the "required" features for 4K Blu-ray are 2160p, 4:2:0 in 10-bit and up to 60 fps (Version 1), with higher bitdepths being an optional feature for players. I'm not sure how they want to do optional support for something like this though, unless they ship a second disc or a second video track on the same disc. Previously, HD audio was optional, and there always has been a non-HD fallback mechanism in place. So this optional higher bitdepth should still be taken with a grain of salt.

SHVC is probably the most straightforward solution. One of the Panasonic UHD Bluray slides (http://av.watch.impress.co.jp/img/avw/docs/683/374/html/dg11.jpg.html) explicitly mentions that the mandatory spec includes HDR in a single independent layer, so I assume that the opposite (HDR in an extra dependent layer) will be an option and that this will extend to other variants of scalability. 12 bit 4:4:4 HDR, please!

[nevcairiel]

You don't really need any extra layer for HDR, you can do it with just metadata in SEI messages, so I wouldn't hold my breath.
Even your chart claims that HDR is only 10-bit, and there is no mention of improved color sampling either. Although SHVC is a possibility, I'm not seeing it yet.

[wanezhiling]

The hevc performance of GM206 is impressive.
But I guess Skylake will beat it easily as Intel is an expert on hardware decoding. ^^

[nevcairiel]

The hevc performance of GM206 is impressive.
But I guess Skylake will beat it easily as Intel is an expert on hardware decoding. ^^

Too bad their drivers tend to suck, causing crashes in things that NVIDIA just works on =)
Besides, on H.264 Maxwell has been equally fast as Intels decoder now.

[Gravitator]

The GeForce GTX 970 can not work with 4 GB of memory efficient manner > official forum (https://forums.geforce.com/default/topic/803518/geforce-900-series/gtx-970-3-5gb-vram-issue/1/)
It is better to wait for the end of the investigation before buying :)

Nai's Benchmark (to analyze the speed of memory sections) > from this topic (http://www.computerbase.de/forum/showthread.php?t=1435408)
--Use at your own risk--

[Stereodude]

Unlikely. UHd Blu-Ray is pretty much confirmed 10 bit 4:2:0 and to be on the market at the end of 2015 while >10 bit HEVC is not finalized yet.
A lot of the rumors regarding what UHD Blu-Ray will do can't be met with with V1. That said, it's anyone's guess what will actually materialize. They'll probably make some stuff optional and the whole thing will be one giant mess by the time you factor in the capabilities of your display, capabilities of your HDMI connection (depending on revision), what the studios decide to put on any particular disc, and what optional features your player can support.

[NikosD]

Too bad their drivers tend to suck, causing crashes in things that NVIDIA just works on =)


The quality of Intel's drivers has been improved dramatically the last few years, although it's not at the same level of Nvidia and AMD yet.

About crashes, if you mean 10bit HEVC hybrid decoding, that was a bug of Nvidia too.


Besides, on H.264 Maxwell has been equally fast as Intels decoder now.

There are 3 different architectures of Maxwell, if we call 960 a 3rd generation, at least regarding media capabilities.

Maxwell 1 and 2, like Nvidia 750 and your 980, I think have exactly the same decoder VP6 which is capable of H.264 4K80 fps.

Haswell has a 4K 120fps H.264 decoder, so it's definitely faster than VP6.

About the new VP7 of 960, I would definitely like to know if H.264 performance has been improved further.

[cyberbeing]

NVIDIA GTX 960's 8bit HEVC decode performance seems as expected. I just hope NVIDIA breaks their habit of limiting their latest and greatest ASIC decode chips to mainstream and entry level GPUs. It would be nice to see something like a GTX 960 Ti on a cut down GM204 with this decode chip, but I'm not getting my hopes up.

The GeForce GTX 970 can not work with 4 GB of memory efficient manner > official forum (https://forums.geforce.com/default/topic/803518/geforce-900-series/gtx-970-3-5gb-vram-issue/1/)
It is better to wait for the end of the investigation before buying :)

Nai's Benchmark (to analyze the speed of memory sections) > from this topic (http://www.computerbase.de/forum/showthread.php?t=1435408)
--Use at your own risk--
FWIW, my GTX 770 Kepler GPU has a similar performance drop-off beyond ~1.6GB-1.7GB/2GB with that Nai's Benchmark, but only when Desktop Compostion is enabled. That particular issue has existed for quite awhile on other GPUs beyond ~80% memory utilization, but maybe this GTX 970 issue is something different. IMHO, it's more interesting that the GTX 980 is supposedly unaffected for reasons unknown. The investigation will be interesting either way, though it's not something which would be worth holding off buying for when the performance and value of the GTX 970 has already proven itself. Though honestly, I personally was holding off on both the GTX 970/980 for the sole reason that they came too early and were lacking the 4K 10bit HEVC decode ASIC which the GTX 960 now has.

Update: http://www.pcper.com/news/Graphics-Cards/NVIDIA-Responds-GTX-970-35GB-Memory-Issue https://forums.geforce.com/default/topic/803518/geforce-900-series/gtx-970-3-5gb-vram-issue/post/4432672/#4432672
NVIDIA has now made a statement the behavior on the GTX 970 is a driver optimization which partitions memory to optimize hardware resource access to the memory subsystem.

[NikosD]

NVIDIA GTX 960's 8bit HEVC decode performance seems as expected.


Well, this time it looks like they did it right, we only have to wait for Nevcairiel to build the 10bit decoder and see the performance on that.

Based on the previous first generation 4K H.264 decoder (VP5), they didn't manage to even reach 4K30fps.
VP5 is a 4K26-28fps decoder.

So, we didn't know exactly what to expect from Nvidia.

[wanezhiling]

Kepler (600, 700 series)'s VP5 can easily reach 4K35fps~4K40fps.
Only GT520/610(GF119) can't reach 4K30fps, even GT520M(GF119 as well) can reach 4K31fps, my friend tested before.

[wizziwig]

Too bad their drivers tend to suck, causing crashes in things that NVIDIA just works on =)
Besides, on H.264 Maxwell has been equally fast as Intels decoder now.

Have you tested the refresh rate accuracy for 23.976 and 59.940 fps on the 960? That seems to be an area where Intel has the edge over everyone else. On Haswell, it just works out-of-the-box without having to mess with custom out-of-spec timings that may not work on all displays.

[nevcairiel]

I've been working on 10-bit decoding, and while it mostly works, it doesn't seem to be likely that it'll work in DXVA2 native mode anytime soon, as EVR just either crashes or refuses to render anything when receiving P010.

The good news is that as a separate side-project I've been working on reducing the CPU usage of Copy-Back, and assuming a direct decoder -> output chain (no software deint, no format conversion, clean nv12 decode and nv12 output, or p010 decode and p010 output), I could reduce the CPU usage on my NVIDIA system by up to 50%, and additionally the benchmarked performance is now 99% the same as DXVA2-Native benchmarks. These gains may be a bit smaller on Intel or AMD GPUs, but they will still be quite noticeable.

I may also add a P010 decode -> NV12 output path using these optimizations, as that may be a common use-case for now when viewing 10-bit content on many renderers.

[Qaq]

The good news is that as a separate side-project I've been working on reducing the CPU usage of Copy-Back, ... I could reduce the CPU usage on my NVIDIA system by up to 50%, and additionally the benchmarked performance is now 99% the same as DXVA2-Native benchmarks. These gains may be a bit smaller on Intel or AMD GPUs, but they will still be quite noticeable.Impressive! That is what I was asking for (http://forum.doom9.org/showthread.php?p=1576230#post1576230). I feel like I'm ready to watch some high bitrate telecined movies now with my Athlon 64 X2 6000+. :thanks:

[nevcairiel]

The Athlon 64 is probably still rather slow for CB, as it doesn't have SSE4.1, which introduced the optimized instructions for copying from GPU memory to system memory. It can make a world of difference.

[Qaq]

Oh no.... we'll see.

[vivan]

I've been working on 10-bit decoding, and while it mostly works, it doesn't seem to be likely that it'll work in DXVA2 native mode anytime soon, as EVR just either crashes or refuses to render anything when receiving P010.What about madVR? It should support DXVA2-native.

[nevcairiel]

What about madVR? It should support DXVA2-native.

I won't bother with the complexity of 10-bit through DXVA2-Native for madVR only, which works much better with CB anyway (or even does CB itself!), not to mention that it probably doesn't accept it right now either.

[NikosD]

I've been working on 10-bit decoding, and while it mostly works, it doesn't seem to be likely that it'll work in DXVA2 native mode anytime soon, as EVR just either crashes or refuses to render anything when receiving P010.


So, it seems that DXVA2 Native and 10bit HEVC decoding are incompatible.

What about EVR-CP ?
Any difference ?

Could Intel or AMD do any tricks with the driver, in order for DXVA2 Native and 10bit to be compatible ?

[huhn]

So, it seems that DXVA2 Native and 10bit HEVC decoding are incompatible.

What about EVR-CP ?
Any difference ?

Could Intel or AMD do any tricks with the driver, in order for DXVA2 Native and 10bit to be compatible ?

the problem is that current renderer can't handle P010 in general.
maybe the current or next version of madVR will handle P010 DXVA input correctly.
EVR can't even handle P010 from software decoding...

[NikosD]

Then Microsoft should build a better renderer for Windows 10 in order to use 10bit.

I don't use madVR.

[huhn]

to display 10 bit videos in 10 bit you need a directx 10+ fullscreen surface too a lot needs to be changed.

[mhourousha]

So, it seems that DXVA2 Native and 10bit HEVC decoding are incompatible.
What about EVR-CP ?
Any difference ?
Could Intel or AMD do any tricks with the driver, in order for DXVA2 Native and 10bit to be compatible ?
GPU can't handle any YUV420 format(NV12 P010 P016 etc) directly before D3D11.1.
therefore if you want avoid cpu copy-back,you need convert YUV 420 to RGB(StretchRect or VideoProcessBlt)in DXVA pass, in render pass,convert it back to YUV420,apply some upsample and color-space mapping algorithm,or use the RGB surface directly.
the problem is most GPU can't do P010/P016->RGB conversion under D3D9EX,so if decoder output P010/P016 surface，then can't convert it to D3D compatible format with GPU.
D3D11.1 can read YUV420 format directly(mapping to 2 SRV one for Y and the other for UV)。so it no need to do the YUV->RGB conversion in DXVA pass。but it need a DX11 Decoder(Decoder use DX11 device instead of D3D9Ex)。D3D9Ex Device can't share YUV420 DXVA Surface to DXGI_FORMAT_NV12 or DXGI_FORMAT_P010/DXGI_FORMAT_P016

[NikosD]

Microsoft's MFT decoders are all D3D11, but I think LAV Video is D3D9.

[nevcairiel]

DirectShow EVR is D3D9 only, it doesn't offer a D3D11 interface to DXVA2. No experience with MediaFoundation to say if thats any different.
Its possible that MF EVR is a lot better and supports all we need already, but other than Windows Media Player, nothing uses MF.

Microsoft is unfortunately not going to care about DirectShow in Windows 10 anymore, they haven't really cared about it for a long time now.

[huhn]

i guess they want to "force" developer to switch to MF.
at least they start to add some small reasons to do this.

[cyberbeing]

On my GTX 770 (PCI-E 3.0 x8) & i5-3570K @4.4Ghz, the new LAV dxva2cb-Direct mode only has a minor performance benefit with the 4K HEVC hybrid decoder, but it does reduce CPU usage somewhat. dxva2native continues to be much faster with the HEVC 4K hybrid decoder on this GPU.

3.Ducks-2160p@50fps-4Mbps
DXVAChecker x64 - Playback Benchmark (scaled to 1280x720)
DXVA2 Native: 187.350 fps | CPU Usage 28% | GPU Load ~93%
DXVA2 Copyback: 66.561 fps | CPU Usage 21% | GPU Load ~52%
DXVA2 Copyback Direct: 73.253 fps | CPU Usage 15% | GPU Load ~52%

[nevcairiel]

The hybrid decoder is CPU bound, which is why it behaves quite differently. Additionally, it probably has to upload a lot more data to the GPU itself, which I then download again, and upload again, which would explain how its generally not ideal for a CB workflow.
On H.264, or using my 960 with HEVC/HEVC 10, the speed difference is practically non-existent with minimal CPU usage (< 2%, down from 5-6%) now.

This mode does allow however using CB for HEVC 4K Main10 files on the 960 without any significant performance impact, while DXVA2-Native for 10-bit just doesn't work with the software we have.

[cyberbeing]

I figured as much, which is why I've always seen the hybrid decoder as nothing more than a crutch. If copyback-direct greatly improves performance using the modern ASIC decoder on your GTX 960, that's all that really matters to me. What sort of decoding performance discrepancy were you seeing between 'normal dxva2 copyback' and 'dxva native' on your GTX 960 prior to these changes?

The ASIC decoder on this GTX 770 seems too slow to bottleneck copyback on high-bitrate 4K H264. Testing a 100Mbps clip, the biggest difference was lower CPU usage:

DXVAChecker x64 - Playback Benchmark (scaled to 1280x720)
37.342fps | 0% CPU (Native)
37.206fps | 3% CPU (Copyback-Direct)
37.198fps | 5% CPU (Copyback)


On something like 10Mbps 1080p H264 baseline clip used by WinSAT, again it mainly saved CPU time:

DXVAChecker x64 - Playback Benchmark (scaled to 1280x720)
175.522 fps | 0% CPU (Native)
166.549 fps | 5% CPU (Copyback-Direct)
163.770 fps | 8% CPU (Copyback)


That said, I can't complain about lower CPU usage with slightly higher performance even if DXVA2 Native retains an edge. :)

[NikosD]

The hybrid decoder is CPU bound, which is why it behaves quite differently.


Using a Core i7-4790 with an iGPU HD 4600, it's definitely GPU bound.


This mode does allow however using CB for HEVC 4K Main10 files on the 960 without any significant performance impact, while DXVA2-Native for 10-bit just doesn't work with the software we have.

Can you post some figures in Playback Performance mode for 10bit 4K HEVC in DXVA Copy-Back ?

For example the ASTRA clip or any other demanding clip which mimics UHD Bluray.

[nevcairiel]

On the Astra clip, using DXVA2 Copy-Back in direct mode. In the meantime I implemented Direct Mode for P010 out, and a direct mode for P010 decode -> NV12 out (for renderers that don't do 10-bit)

Decode, Direct P010 Out: 127 fps, 3% CPU
Playback (1280x720), Direct P010 Out: --- EVR doesn't do P010 out.

Decode, Direct NV12 Out: 126 fps, 3% CPU
Playback (1280x720), Direct NV12 Out: 116 fps, 6% CPU

And for giggles without the new Direct Mode:
Decode, No Direct P010 Out: 73 fps, 6% CPU

A significant performance loss due to the extra copying being done right there.

[NikosD]

So, it's a 4K 120fps decoder even for 10bit in DXVA-CBD.

Impressive.

Does the HW decoder stutter in those Samsung 4K 10bit HEVC files like ffmpeg SW decoder ?

[nevcairiel]

That stuttering thing should have been fixed a long time ago already.
At least the "UHD_ENT_Transformer_Quad.ts" that I have, which is a Samsung 10-bit clip, plays perfectly in direct mode (or in SW mode for that matter).

[NikosD]

Well, I had tried your fix which worked fine, but you told me that it wasn't an official fix, but rather a "fix", like a workaround.

Back then, FFMpeg developers hadn't included it in their updates.

I don't know if recent FFMpeg updates include your "fix" or an official fix :)

But, yes it works.

[prxy99999]

I've been evaluating hevc decoders for 1080p playback on older hardware like core2duos and core2quads. I've found the lav filters (both 32 and 64bit) indistinguishable visually on playback on the core2quads from a decoder like strongene. In the case of the core2duos however when it comes to a scene where the camera pans across from side to side or up and down, especially if there is a lot of action in the scene than I do get lots of stutter and delay with lav but the strongene decoder plays them smoothly.

According to the strongene webpage and manual the decoder is free for personal use. But I'm finding on all systems that as soon as any clip hits the 22-23 min mark the picture will freeze and remain static while audio continues. If I stop the clip and seek to just before it freezes and continue playback it continues on for another 22-23mins. I've tried searching the web and emailing them but haven't had a reply. So was wondering if anyone else knows whether the strongene decoder is free to use but "time limited" for playback.

The version of the strongene CPU decoder i'm using is the one currently available for download on the website:

Strongene H.265/HEVC Decoder for Windows
Version:2.0.2.6 Release: 2014/06/10

[Neet009]

CPU:core2quad Q9550
GPU:GTX 960 (driver 347.25)
OS:win7 x64
I tried LAVFilters-0.63.0.74-git, it seems DXVA2 Copy-Back direct mode doesn't work with 10bit HEVC clips, and return to use SW decode. I tested ASTRA clips and Samsung 4K 10bit HEVC clips with EVR and madVR, all of them doing the same result.
http://i.imgur.com/OXL7RYq.png

However, when playing 8bit HEVC clips, DXVA2 Copy-Back direct mode works normally as expected.
http://i.imgur.com/w1QccVo.png

Use DXVAchecker to check 10bit HEVC clips, it show this:
http://i.imgur.com/5GL6wsV.png
http://i.imgur.com/Yo747cM.png
http://i.imgur.com/athd4N5.png

8bit HEVC clips:
http://i.imgur.com/0mii44h.png
http://i.imgur.com/meTRPIv.png

Did l do something wrong?:confused:

[nevcairiel]

10-bit is not supported yet in the nightly builds.

[Neet009]

10-bit is not supported yet in the nightly builds.

I got it. Thank you for reply.

[NikosD]

According to the strongene webpage and manual the decoder is free for personal use. But I'm finding on all systems that as soon as any clip hits the 22-23 min mark the picture will freeze and remain static while audio continues. If I stop the clip and seek to just before it freezes and continue playback it continues on for another 22-23mins. I've tried searching the web and emailing them but haven't had a reply. So was wondering if anyone else knows whether the strongene decoder is free to use but "time limited" for playback.

The version of the strongene CPU decoder i'm using is the one currently available for download on the website:

Strongene H.265/HEVC Decoder for Windows
Version:2.0.2.6 Release: 2014/06/10

Using the same latest version of Strongene HEVC decoder and DXVA Checker as media player, I had no problem of decoding a HEVC movie up to 40min and then I stopped.

So, it looks like a problem in your system/media player rather than the Strongene decoder.

[NikosD]

On the Astra clip, using DXVA2 Copy-Back in direct mode.

Could you test this small (92MB) and short (25s) clip with your 960 DXVA-CBD decoder ?
http://www.filedropper.com/hotbird4kdemo

It's a 10bit 2160p50fps HEVC clip with 30Mbps bitrate that I can't play in realtime even with plain EVR and latest LAV x64 nightly version.

[v0lt]

On the Astra clip, using DXVA2 Copy-Back in direct mode. In the meantime I implemented Direct Mode for P010 out, and a direct mode for P010 decode -> NV12 out (for renderers that don't do 10-bit)
It is possible to convert to YUY2 or directly into RGB32?

[nevcairiel]

It is possible to convert to YUY2 or directly into RGB32?

No, that would defeat the primary purpose of direct mode (speed), since those conversions are very slow.
It also wouldn't be very direct anymore. :)

Especially with 4K content, and even more so 10-bit, its crucial that we don't need or want to perform software processing on the video, because its going to be super slow.
Any processing should be done on the GPU, so direct mode will not offer any conversions that are not absolutely required to get the image to the GPU. In short, this is NV12 -> NV12 out, P010 -> P010 out, and as one special case, P010 -> NV12 out, since not every renderer accepts 10-bit (and this kind of conversion is very fast)

I may add NV12/P010 -> YV12 as well because its a rather simple conversion, but I have not decided yet if its really all that useful. Anything that would benefit from it already adds a high overhead, so its kind of meh.

[nevcairiel]

Could you test this small (92MB) and short (25s) clip with your 960 DXVA-CBD decoder ?
http://www.filedropper.com/hotbird4kdemo

It's a 10bit 2160p50fps HEVC clip with 30Mbps bitrate that I can't play in realtime even with plain EVR and latest LAV x64 nightly version.

Decode to P010, 120 fps (116-125)
Decode to NV12, 117 fps (105-121)
Playback in NV12, 1280x720, 107 fps (66-113)

All tests in direct mode of course.

[NikosD]

Decode to P010, 120 fps (116-125)
Decode to NV12, 117 fps (105-121)
Playback in NV12, 1280x720, 107 fps (66-113)

All tests in direct mode of course.
Thanks for testing.

I have a few more general questions.

For most of us with 8bit displays (TV screens, PC monitors), what's the difference of a 10bit vs 8bit clip in terms of picture quality.

Can we actually see any difference on our displays due to 10bit encoding ?

What's the pipeline of 10bit decoding and displaying a 10bit clip on a 8bit monitor and which component does the conversion from 10bit to 8bit and how ?

The decoder, renderer, GPU driver ?

Are there many ways for 10bit to 8bit conversion ?

[clsid]

The primary benefit of 10bit is that the encoder can work with a higher precision. This results in better compression and/or better quality. So 10bit encodes can actually be smaller than 8bit ones, and it also benefits quality, particularly giving less banding. These benefits are there regardless of whether you end up decoding to 8bit or not! The potential larger range of colors that is available if outputting 10bit is much much much less important.

So yes, 10bit has benefits, even for 8bit displays. In fact, some displays even use less than 8bits internally.

Most video renderers only support 8bit input. In that case the conversion is done by LAV Video decoder. It uses high quality conversion, with dithering.
MadVR is the only DirectShow renderer that supports 10bit input. In that case, conversion is done by MadVR. Again, high quality with dithering.

[NikosD]

Thanks for the prompt and detailed reply.

[Neet009]

CPU:core2quad Q9550
GPU:GTX 960 (driver 347.25)
OS:win7 x64

I did some test with LAV 0.63.0.75-git:
Astra clips(4K 10bit HEVC), Copy-Back direct mode
DXVAChecker x64, decode, P010 out
76~78 fps , 15% CPU
DXVAChecker x64, decode, NV12 out
92~93 fps, 14% CPU

DXVAChecker x86, decode, P010 out
67~69 fps, 15% CPU
DXVAChecker x86, decode, NV12 out
86~88 fps, 14% CPU

Ducks-2160p@50fps-4Mbps, Copy-Back direct mode
DXVAChecker x64, decode
152~154 fps, 15% CPU

DXVAChecker x86, decode
153~154 fps, 16% CPU

It seems the 10bit results are slower than Nevcairiel's test. What reason is likely to make a difference? Is my cpu too old or lack of some instruction sets?

[cyberbeing]

It seems the 10bit results are slower than Nevcairiel's test. What reason is likely to make a difference? Is my cpu too old or lack of some instruction sets?

Assuming nevcairiel's i7-4770K results still hold true, it's probably both your CPU & Memory subsystem being too slow, especially if you see a similar performance difference with P010 vs NV12 output when using software decoding of 10bit H264 video. P010 output requires transferring 16bit (padded) data, while NV12 is only 8bit data. Use of P010 output doubles the bandwidth requirements, the amount of data being transferred across system memory, and to/from your GPU over PCI-E. It's not abnormal for the overhead of outputting P010 directly to be larger than the CPU overhead of converting to NV12, especially on an older system.

[nevcairiel]

Note that I'm on a i7-5930K now with DDR4 memory, so my memory performance is quite good, and core 2 is quite old, the memory may be much more of a bottleneck there than on newer CPUs.

Interesting is the difference between NV12 and P010. LAV has to read the full P010 from the GPU in any case, in the NV12 case its just converted before writing to sysmem (in processor registers), so that would clearly point to the system memory being a bottleneck here, since the only major difference is the amount of memory being written into system memory.

[Neet009]

OK, it seems that system memory would be bottleneck on some old systems when using 4K 10bit HW decoding. Thank you for explaining.

[NikosD]

CPU:core2quad Q9550
GPU:GTX 960 (driver 347.25)
OS:win7 x64


Could you test using your GTX 960 with latest DXVA Checker 3.3.2 x64 in DXVA native and "decode" mode using latest LAV x64 0.63.0.75 the clip below ?

http://xhmikosr.1f0.de/samples/2160p/DucksTakeOff/DucksTakeOff_2160p50.x264.CRF25.mkv


It's a 2160p H.264 video with huge bitrate.

[Neet009]

Could you test using your GTX 960 with latest DXVA Checker 3.3.2 x64 in DXVA native and "decode" mode using latest LAV x64 0.63.0.75 the clip below ?


OK!
DucksTakeOff_2160p50.x264.CRF25
DXVAChecker 3.3.2 x64, DXVA native, decode
78.5 fps, cpu 2% (Test 5 times, Avg)

DXVAChecker 3.3.2 x64, DXVA copy-back direct, decode
77.9 fps, cpu 9%

[NikosD]

OK!
DucksTakeOff_2160p50.x264.CRF25
DXVAChecker 3.3.2 x64, DXVA native, decode
78.5 fps, cpu 2% (Test 5 times, Avg)

DXVAChecker 3.3.2 x64, DXVA copy-back direct, decode
77.9 fps, cpu 9%

Thanks! I thought so!

It seems that Nvidia uses the same H.264 decoder for 960, like the previous models VP6.

They didn't improve H.264 performance, so it's still a 4K80 fps H.264 decoder.

Just for comparison, my Core i7-4790- iGPU HD 4600 (GT2) at 1.5GHz has these results:

DucksTakeOff_2160p50.x264.CRF25
DXVAChecker 3.3.2 x64, DXVA native, decode
200 fps, cpu 1% (Test 5 times, Avg)

[NikosD]

CPU:core2quad Q9550
GPU:GTX 960 (driver 347.25)
OS:win7 x64


NVIDIA GTX 960 Benchmarks


I encoded a 2160p50fps-HEVC_Main10 file with an average bitrate of 128Mbps close to BluRay UHD.

The clip is here:
http://www.filedropper.com/crowdrun2160p50main10-127mbps

I also added, just for testing reasons, a 300Mbps clip here:
http://www.filedropper.com/crowdrun2160p50-hevcmain10300mbps

I would like to see some results using LAV x64 v0.64 DXVA-CBD of Nvidia 960 of both decode and playback modes.

I'm interested in MIN/AVG/MAX fps.

Using my Core i7-4790 the results are:

1st clip (130Mbps)
Decode: 20/38/42
Playback (1280x720):12/33/35

2nd clip (300Mbps)
Decode: 3/27/30
Playback (1280x720):0/24/28

[nevcairiel]

Decode (P010): 94/100/104
Decode (NV12): 94/100/105
Playback (NV12, 720p): 36/89/101

I'm not sure when the Playback min value occured, it did seem to run through the video very fast at all times. Maybe right at the start when it was still loading or something. Not sure I would put too much value on it.

[NikosD]

Decode (P010): 94/100/104
Decode (NV12): 94/100/105
Playback (NV12, 720p): 36/89/101

I'm not sure when the Playback min value occured, it did seem to run through the video very fast at all times. Maybe right at the start when it was still loading or something. Not sure I would put too much value on it.

Just added a 300Mbps test clip.

It's either that you write or the clip has some spikes with huge bitrate.

A simple bitrate viewer I have it's not compatible with HEVC in order to check it out.

[Neet009]

My results:
(GTX 960, Q9550 3.4GHz, DDR3-1333 16G)
LAV x64 v0.64 DXVA-CBD

1st clip (130Mbps)
Decode(P010): 73/80/83
Decode(NV12): 84/92/95
Playback (1280x720):54/85/91

2nd clip (300Mbps)
Decode(P010): 60/78/82
Decode(NV12): 64/79/84
Playback (1280x720):39/75/84

[P.J]

LAV 0.64.0 got much better but the only problem is the glitches :(
The video doesn't play smoothly while the CPU/GPU are ~50%

[NikosD]

For anyone interested in testing, I encoded a 8K (7680x4320) HEVC clip at 60fps in both 10bit and 8bit versions.

Of course, you can't HW accelerate the decoding of those clips even if you have a 960 card, due to high resolution.

You need a very fast CPU.

The clips are just 1 sec and very small.

Clip 1- HEVC 4320p60fps -10bit
http://www.filedropper.com/crysis3-7680x432060fps-137mbps1sec-hevc10bit

Clip 2 - HEVC 4320p60fps -8bit
http://www.filedropper.com/crysis3-7680x432060fps-141mbps1sec-hevc


Tests

DXVA Checker x64 3.3.2 - LAV x64 v0.64 - Core i7 4790


Clip 1

Decode 2/16/17


Clip 2

Decode 10/21/22


Note that I'm on a i7-5930K now with DDR4 memory, so my memory performance is quite good, and core 2 is quite old, the memory may be much more of a bottleneck there than on newer CPUs.


I'm sure you are going to need a faster processor, like Core i7-5960X for realtime decoding of those 8K HEVC clips :)

[foxyshadis]

What was the cpu utilization like? Pegging all 8 threads? I wonder how much of that is just CABAC processing at 140mbps.

[NikosD]

It sure uses all 8 threads and the CPU utilization is on average ~80% by putting 16 threads in LAV Video properties, as in the above benchmark tests.

Looking at the decoding performance of the 4K HEVC 10bit ~130Mbps, it has more than double fps (38 fps vs 16 fps) than 8K HEVC 10bit ~140Mbps.

So, the resolution plays the most important role for decoding.

[Aleksoid1978]

Hi all.
I buy GTX 960, test LAV Video Decoder on HEVC 10bit, then modify MPC-BE's video decoder. And ... perfect work DXVA HEVC 10bit, EVR Custom accept P010 DXVA input.
Here test build - http://aleksoid.voserver.net/MPC-BE/mpc-be.exe_DXVA_HEVC_4K_10Bit.7z - it's test build and work only 10bit HEVC DXVA decoder.

P.S. Only MPC-BE' EVR Custom can accpet P010 DXVA.

[NikosD]

P.S. Only MPC-BE' EVR Custom can accpet P010 DXVA.

So, does that mean that you can use DXVA native with 10bit HEVC ?

I asked nevcairiel about EVR-CP but he didn't answer.

What's the difference between MPC-HC EVR custom and MPC-BE EVR custom ?

[Aleksoid1978]

So, does that mean that you can use DXVA native with 10bit HEVC ?

What's the difference between MPC-HC EVR custom and MPC-BE EVR custom ?

Yes - native DXVA HEVC 10bit but only in EVR Custom, vanilla EVR show only white screen.

About MPC-HC - it's use hook and drop P010 input for EVR :)

[wanezhiling]

http://i2.tietuku.com/91a1f3f716deb554.gif

Any proof?
Like dxva checker trace log.

[Aleksoid1978]

Proof - 1.5-3% cpu load when decoding hevc 10bit 4k :)

[NikosD]

This is my post asking nevcairiel about EVR-CP.

He didn't answer, but probably he didn't thought about MPC-BE.

So, it seems that DXVA2 Native and 10bit HEVC decoding are incompatible.

What about EVR-CP ?
Any difference ?

Could Intel or AMD do any tricks with the driver, in order for DXVA2 Native and 10bit to be compatible ?

[Aleksoid1978]

What interesting - HEVC DXVA 10bit decoder support output to "standart" NV12 type.

http://i.imgur.com/XQhZjgd.png

http://i.imgur.com/0sAiMMD.png

[nevcairiel]

As long as vanilla EVR does blindly fail on it, i'm not enabling it, its as simple as that. I don't add buggy features that only work in one player with one specific renderer, and crap out everywhere else.
And I'm definitely not enabling NV12 output from the decoder without proper dithering... :)

Let us know when you get EVR to actually accept a P010 connection, and not NV12, then you got something new. :p Your screenshot has 8-bit surfaces everywhere, thats no good for quality.
Using a NV12 connection and sending P010 DXVA2 surfaces is something I tried before as well, but it crashes on some systems, shows white screens on others...

[Aleksoid1978]

Let us know when you get EVR to actually accept a P010 connection, and not NV12, then you got something new. :p
Using a NV12 connection and sending P010 DXVA2 surfaces is something I tried before as well, but it crashes on some systems, shows white screens on others...

Only EVR Custom/EVR Sync accept NV12 for HEVC 10bit. Vanilla EVR - show white screen, madVR - green :)

[wanezhiling]

EVR-CP works, that's enough, most people use it cuz it's a default renderer in most third-party win players. :)
Well, how to do a proper dither, it's a problem..

[NikosD]

Plain EVR doesn't even support external subtitles.

EVR custom is heavier but more useful in realworld situations.

[Aleksoid1978]

EVR Custom accept input P010 and do output to RGB32(mixer X8R8G9B8):
http://i.imgur.com/Tx2NK35.png

[P.J]

Just ordered my EVGA GTX 960, can't wait :D

[NikosD]

Carrizo from AMD will be the first CPU in mid 2015 with a full fixed-function H.265 decoder, but it seems it's for mobile only. No desktop Carrizo.
According to AMD, it has more than 3.5x transcode performance than Kaveri.
http://www.anandtech.com/show/8995/amd-at-isscc-2015-carrizo-and-excavator-details

[nevcairiel]

Technically, the decoder is part of the iGPU, not the CPU (albeit both part of the same chip of course). And who knows if Skylake beats AMD to it... :)

[NikosD]

Almost all CPUs these days have iGPUs inside and almost all iGPUs have fixed - function transcoders inside too.

So a CPU is actually a triple chip (CPU/GPU/Transcoder)

Carrizo is relevant because it will be used in laptops and notebooks etc and I doubt that Skylake will be available before Q4 if it catches 2015 :)

[STaRGaZeR]

Carrizo is not relevant at all, it's probably built into cheap, nasty notebooks with cheap build quality & 1366x768 TN panel screens. Of course, if thats what you want to buy, go ahead.

You don't know where Carrizo will be, so stop spreading BS.

[NikosD]

Easy guys...

GTPVHD is an Intel fan and he likes to be over optimistic for Intel and aggressive to AMD, but let's not continue this topic in this thread.

[StinDaWg]

Does anyone know if Nvidia's GTX 960 H.265 decoder works with dxva copyback and cuvid? I saw a review where they were using it with dxva2, but haven't seen if it's compatible with the other modes.

[P.J]

Finally got my GTX960, no more glitch like the Hybrid solution!
Anyway to play 10bit HEVC with LAV/MPC-HC?

[nevcairiel]

LAV supports 10-bit HEVC decoding on the GTX 960 just fine, just use DXVA2-CopyBack and it'll work.

[iSunrise]

What's the highest HEVC profile/level GM206/GTX 960 does support? Level 5.1?

[NikosD]

Level 5.1 is the old max of H.264, now it's 5.2.

For HEVC maximum level is 6.2 and should be supported by GTX 960

[iSunrise]

Level 5.1 is the old max of H.264, now it's 5.2.

For HEVC maximum level is 6.2 and should be supported by GTX 960
Yes, but is it actually the case? Especially Level 6.2 seems way too high. I would expect Level 5.2 at the most.

Can someone with a GTX 960 verify?

[sneaker_ger]

Just look at the benchmarks that have been posted: GTX960 does not fully support 6.2.

[NikosD]

I think it supports 4K L6.2, what benchmark do you mean ?

Is there any clip not playable ?

[sneaker_ger]

Level 6.2 allows 4K @ 300 fps, 8K @ 120 fps.

[nevcairiel]

5.2 may barely work depending on the complexity, but its only constrained by speed there, not any other features.
6 and up support higher resolutions (8K), which are not supported, so those levels are not supported.

A level is only supported if its fully supported, including speed and resolution constraints. 5.1 should be fully supported, 5.2 might be close on the speed constraint.

Adding extra conditions like "4K at 24 fps may work on L6.2" is just not something thats a useful statement. :p

[NikosD]

I'm on the opposite direction on this.

Sandybridge has a 1080p H.264 decoder capable of 1080p L5.2@300Mbps@100fps.

And because H.264 L5.2 supports up to 4K, does that mean that Sandy doesn't support H.264 L5.2 ?

NO. Definitely not in my opinion.
I could say it's more than capable of 1080p H.264 L5.2 decoding.

So, if GTX 960 can decode a 4K L6.2. 300Mbps at 80fps in DXVA copy-back, I could say it supports L6.2. with the resolution limit of 4K of course.

[nevcairiel]

The point is that these levels are designed to classify a decoder with one simple statement. Classifications are not exact, they are broad categories, so they don't tell you the entire picture, but they do tell you the *minimum* you can expect.
If I wanted to know exactly which limits it has, I would dig up a spec sheet, or run tests, or something.

The only classification valid for the GTX 960 decoder is Level 5.1, since none of the other levels are 100% supported.
This is also how the HEVC standard specifies these levels (and H.264 too). To quote: "A decoder that conforms to a given tier/level is required to be capable of decoding all bitstreams that are encoded for that tier/level and for all lower tiers/levels."

There is a definition so that everyone talking about it will talk about the same thing. Talking about anything else makes no sense, as its confusing at best and at worst mis-represents the facts.

Of course this doesn't mean it cannot decode a select assortment of streams using a higher level, and it can, but there is no classification to express that.
Strictly speaking the old H264 decoders are only L4.1, but that does not mean that they cannot decode 5.1/5.2 1080p streams. But they are not conforming to a full 5.2 decoder!

What you want is something that expresses their full potential, but something like that doesn't exist. The levels only express the minimum features it supports 100%, not the maximum.

PS:
"L6.2 with resolution limit to 4K" also is not correct, as L6.2 allows 300 fps at 4K, and it certainly isn't capable of that.
Do you see how you would need to limit this further and further? Might as well drop the level then and say "it can do 4K, 300 mbps at 80 fps", the level wouldn't add any information anymore.

[NikosD]

I can't disagree with you, especially when you refer to strict definitions, but I remember a time around my first posts here, when the H.264 decoders couldn't decode 1080p H.264 L5.1 streams.

What was the main reason ?

They couldn't handle the 16 Ref.

That feature was critical, because they couldn't play at all those clips.

So, a 1080p H.264 decoder which can't decode at all a certain clip due to the lack of an advanced feature of the level (like ref) does certainly not belong to that level.

So, when ATI fixed the driver for UVD2.2 in order to decode 16 Ref clips, they claimed 1080p L5.1 decoding capability, meaning they fixed the 16 Ref limitation which was critical and not the 300Mbps requirement of L5.1 or the 4K resolution.

Using the rules as strict as you do is one point of view, but in many cases a lot of people reading a GTX 960 is a L5.1 H.265 decoder would understand that they couldn't decode a 4K L6.2@60 fps at all, when we both agree that it can.

Update:


Strictly speaking the old H264 decoders are only L4.1, but that does not mean that they cannot decode 5.1/5.2 1080p streams. But they are not conforming to a full 5.2 decoder!


Looking at the definition of H.264 levels, I'm afraid that your claim about old H264 decoders is not true.

The L4.1 asks for H.264 2048x1024@30fps support, that none of the 1080p old H.264 decoders provide.
The same restriction is true for L4.0

So, according to the strict rules you brought to our conversation, the old decoders like VP4 or QuickSync of SandyBridge are only L3.2 decoders (!)

This is both hilarious and ridiculous...It's completely meaningless and waste of time to talk again about strict level definitions.

[iSunrise]

5.2 may barely work depending on the complexity, but its only constrained by speed there, not any other features.
6 and up support higher resolutions (8K), which are not supported, so those levels are not supported.

A level is only supported if its fully supported, including speed and resolution constraints. 5.1 should be fully supported, 5.2 might be close on the speed constraint.
I basically read all the benchmarks of GM206 and after reading the relevant Wikipedia HEVC article, I could only guess that L5.1 was fully supported by GM206, while I was not sure about L5.2. Also, according to one NV slide the X1 SoC can only do HEVC 4K@30fps (which would conform to L5.0). These guesses were based upon the fact that the maximum resolution and fps for a specific Level was fully met.

So, when taking pre-encoded content into account and purely from a real-time playback perspective, L5.1/L5.2 looks like it's sufficient, which is probably also the reason why they (meaning NV, we will have to see about Intel with Skylake) went with that (Ultra HD Blu-Ray). Even for mobile devices, on current or a future SoC, video encoding/decoding 1080p@300fps and 2160p@120fps is more than enough. For encoding or any other usage scenarios, the more speed, the better, obviously.

We really need a table with confirmed specs and user tests (to see the limits) of devices.

PS:
I already did tests in the past with my iPhone 5s, which has limited 4K H.264 decoding capabilities. Up until today, I am not sure what decoding capabilities the iPhone 6 has, because no one really has tested that. And the iPhone is one of the most sold devices today. Apple only lists what they think is necessary for their needs, not the real capabilities of their PowerVR encoder/decoder that is integrated into their SoC (because it's not relevent for the majority).

[NikosD]

From AnandTech GTX960 launch preview:


A full review is expected in the following days.

It's almost two months since the launch preview article about Nvidia GTX 960 in the Anandtech site.

Did they forget to review the card ? :rolleyes:

[iSunrise]

It's almost two months since the launch preview article about Nvidia GTX 960 in the Anandtech site.

Did they forget to review the card ? :rolleyes:
Still no review, unfortunately.

BTW, NV seems to have added H.264 lossless support to their drivers (at least for Linux according to https://devtalk.nvidia.com/default/topic/821171/unix-graphics-announcements-and-news/linux-solaris-and-freebsd-driver-349-12-beta-/). Does this mean that the performance is the same as with the H.264 lossy settings? They really surprise me, seems NV makes huge improvements when it comes to their hardware VDPAU feature-set or drivers respectively.

[P.J]

I have never seen H.264 lossless but H.264/MPEG2 4:2:2

[P.J]

Cyberlink Video Decoder from PowerDVD v14 does support OpenCL H.265 hw-assisted video decoding and it is available to other DS players, it just has a 'do not use' merit.
You also have to set the decoder to use HAM-mode, don't remember if DXVA-mode would also use OpenCL hybrid decoding. When playing a HEVC video go to the Filters-tab, press and hold 'Ctrl'+'Left Mouse' on the 'Cyberlink Video Decoder (PDVD Generic)', it will open advanced/detailed settings view, the Profile column will tell you if the decoder is using OpenCL or not.
It wont use OpenCL on my GTX 550Ti, but does work on my brothers AMD Radeon HD 5700, too bad, because mine does a lot better on all my samples and tests.
But even in SW-mode I think it does even better than Lentoid and it also supports 10-bit, which Lentoid does not, and both are a LOT faster than LAVVideo x86, on my brothers x64-machine also LAVVideo x64 was slower than Cyberlink and Lentoid, of course it has an old AMD Phenom X4 2.6 Ghz with up to SSE4A instructions.

P.S. last LAV build I tested was 0.63.0.12, don't know how much faster 0.63.0.41 is, I did read it should have better error handling now.

No luck with GTX 960 too :rolleyes:

[vivan]

Multicoreware claim that their decoder is "the fastest, most reliable HEVC software decoder in the world". It's now (temporary, lol) available for free https://x265.com/store/index.php/download-free/ (it requires some stupid registration). Also it will screw file associations (since they know better than you that WMP is the best player ever) and maybe mess with merits or other things.
UPD: actually no need to install, just unpack and build graph using UHDcodeSrcFilter.dll and UHDcodeDecFilter.dll filters. But it still will require activation.

Of course this claim is wrong.
Beauty-2160p@30fps-12.3Mbps on i5-4670K:
LAV x64 (16 threads): 85 fps
LAV x64 (auto threads): 68 fps
MS MFT (DXVA): 46 fps
MCW x64: 42 fps

[nevcairiel]

They have been claiming that theirs is the fastest and only complete decoder for quite some time now, even after being informed of the state of the ffmpeg decoder. What can you do. ;)

[NikosD]

I've downloaded and used for a while the first public release of HEVC decoder by MultiCoreWare - UHDcode.

I have to thank Tom (aka x265_Project) from MCW for providing the app and the coupon, which is free right now as vivan wrote above at https://x265.com/store/index.php/checkout/cart/

Before benchmarks, I have some general comments to make:

1) There is no support of 10bit (main10 profile)

2) I couldn't manage to use the UHDcode HEVC plugin with .mkv, .ts or any other container besides .mp4.
So, it seems that right now it can be used only with elementary streams (.265, .h265, .hevc) and .mp4

3) There is no property sheet for the decoder nor for the splitter. So, for example I couldn't change the number of threads, although the decoder used all 8 threads.

4) The only way to actually use the decoder even with .mp4 files is only if you let the UHDcode splitter handle the .mp4 container.

In that way I couldn't enumerate both LAV video and UHDcode at the same time in DXVA Checker.
I had to change the preferred splitter for .mp4 between LAV and UHDcode splitters in order to test both decoders.

5) I tried to install it to my second Win 10 system with a Core 2 Quad processor, but it was already activated in my main system so I couldn't install it there.
It doesn't allow me to register the UHDcodeDecFilter.dll in a second system.

6) After the installation, the .mp4 files are associated with WMP

-----------------------------------------------------------------

All benchmarks were done on a Core i7 4790 with Win x64 8.1 Pro system, using latest DXVA Checker x64 v3.3.2 in Benchmark decode mode.

I used LAV Video x64 v0.64.38 , MPC Video Decoder x64 v1.4.4.265 , both using 16 threads and UHDcode v1.0.5.3/1.0.5.4


1. Girls -1080p60 fps -11Mbps


LAV x64 214/300/396 CPU 70%

MPC x64 216/298/379 CPU 70%

UHDcode 140/163/220 CPU 70%



2.Beauty-2160p@30fps-12.3Mbps

LAV x64 81/94/96 CPU 78%

MPC x64 80/93/96 CPU 78%

UHDcode 36/53/57 CPU 80%


Well, the numbers speak for themselves, I don't have to add anything.
For the same CPU utilization, the performance difference is huge on an AVX2 CPU at least.

P.S

In the package of the installer, it seems that there is an OpenCL component of UHDcode HEVC plugin which wasn't installed on my Intel iGPU system.
I don't know if it can be used with other dGPU cards by AMD and Nvidia.

[vivan]

The UHDcode DirectShow filter supports all video resolutions and frame rates, but as most consumer video displays only support 8 bits/pixel, it is limited to HEVC Main profile. A professional version of UHDcode is available for professional products.What a lame reason.
So they do have 10 bit decoder, but to get it you need to pay even more money.

[x265_Project]

All benchmarks were done on a Core i7 4790 with Win x64 8.1 Pro system, using latest DXVA Checker x64 v3.3.2 in Benchmark decode mode.

I used LAV Video x64 v0.64.38 using 16 threads and UHDcode v1.0.5.3/1.0.5.4

1. Girls -1080p60 fps -11Mbps

LAV x64 203/287/370 CPU 70%
UHDcode 140/163/220 CPU 70%

2.Beauty-2160p@30fps-12.3Mbps

LAV x64 81/94/96 CPU 78%
UHDcode 36/53/57 CPU 80%

Our UHDcode DirectShow filter wasn't designed for benchmarking this way. We've done performance testing against other open-source HEVC decoders, and UHDcode always compared favorably. The DirectShow filter is designed to decode frames and feed them to the player application as requested. It doesn't have a benchmarking mode where it would decode as fast as possible. I'll ask our team to try this out however.

BTW, UHDcode is OpenCL accelerated on the latest AMD discrete and APU integrated graphics, and we will to support Intel and NVIDIA platforms as soon as possible.

[x265_Project]

What a lame reason.
So they do have 10 bit decoder, but to get it you need to pay even more money.
Even more than zero?

[NikosD]

Our UHDcode DirectShow filter wasn't designed for benchmarking this way. We've done performance testing against other open-source HEVC decoders, and UHDcode always compared favorably. The DirectShow filter is designed to decode frames and feed them to the player application as requested. It doesn't have a benchmarking mode where it would decode as fast as possible. I'll ask our team to try this out however.


There is no other way to test the performance of a decoder, but pushing it to the limit without the renderer limitation of realtime playback.

Moreover, I tested the decoder in renderless mode (pure decoding), so there are no limitations at all.

I'm open to suggestions for both method and tools.


BTW, UHDcode is OpenCL accelerated on the latest AMD discrete and APU integrated graphics, and we will to support Intel and NVIDIA platforms as soon as possible.


Nvidia doesn't have a decent OpenCL driver. You have to use CUDA.

For Intel, I managed to run Kernelbinary.exe, IIRC correctly the name and it used OpenCL driver of Intel in order to create some files.

Every other OpenCL decoder even built for AMD (like Lentoid or Cyberlink) can be used by Intel because Intel has a real OpenCL 1.2 driver.

So, you only have to allow it in order to be used by Intel.

[huhn]

what are you talking about? nvidia openCL work totally fine.
and openCL 1.2 is not that important...

[huhn]

i tried to use UHDcode on windows 10 to compare it with the build in HEVC decoder from Microsoft which can do 10 bit BTW.
but DXVA checker got stuck when i tried to play a file with that decoder.

after that started windows media player and watched under plugin if i could see this decoder plugin but nothing to see there. and yes i made sure it was the 64 bit version.

after that i played a file in 32 and 64 bit and didn't notice a huge difference in CPU load.

[NikosD]

what are you talking about? nvidia openCL work totally fine.


What are you talking about ?

What do you mean "Nvidia OpenCL works totally fine" ?

CUDA works fine, not OpenCL.

You are confusing things....

[huhn]

do you have an example?

madVR nnedi3 uses openCL and works totally fine with nvidia.

[NikosD]

Cyberlink's OpenCL HEVC decoder and Lentoid OpenCL HEVC decoder.

There is no OpenCL driver from Nvidia.

All there is, is just a wrapper over CUDA translating OpenCL calls to CUDA calls.

Nvidia doesn't provide any help, maintenance or update to its OpenCL driver.

Intel and AMD have already moved to OpenCL 2.0 and Nvidia is stuck to 1.1 because OpenCL is directly competitive to their proprietary CUDA technology.

Don't expect anything from Nvidia other than just a wrapper over CUDA.

[huhn]

and why is the driver broken now? if i want i can write nirmal programs that doesn't work with intel CPU does that mean intel driver are broken?

if nvidia is doing a hybrid decoder there self they will very likely use cuda no secret.
and they provide this decoder already using CUVID or DXVA on older GPUs.

[huhn]

how about claiming down a bit?

BTW. i can't disagree that a CUDA hybrid decoder has more potential than a openCL hybrid decoder. but this would mean a lot of code needs to be rewritten a little over kill if openCL has to work too and it should be generally the same code for all 3 brands.

and this is no reason to call the nvidia openCL driver broken.

BTW. i was trying to use the UHDcode decoder with my r9 270.

and the 349.90 and 349.65 are windows 10 driver i tried both of them already.

[foxyshadis]

Geez, I come back from a month away and the place explodes the day I get back. Try to keep the discussion to numbers, capabilities, artifacts, and official announcements, not your opinions of each other.

[NikosD]

Looks like Nvidia changes its mind about OpenCL and starting to fix the OpenCL driver.
http://anandtech.com/show/9139/nvidia-posts-35005-hotfix-driver-fixes-games-adds-opencl-12-support

What a coincidence ! Maybe they are reading our posts :D


First announced in late 2011, OpenCL is a point update for OpenCL 1.x that added a handful of new (but potentially important) features. More importantly however is the fact that until now NVIDIA has declined to support OpenCL 1.2, opting instead to direct their energies into the growing CUDA ecosystem and its wider usage than OpenCL 1.x. Consequently the addition of OpenCL 1.2 support at this stage comes as a bit of a surprise, but we’ll take it. And hopefully this is a sign that NVIDIA is going to be catching up on OpenCL so that they can support SPIR/SPIR-V and OpenCL 2.x.


But it remains as a CUDA wrapper.

[nevcairiel]

OpenCL 1.2 has been available in the 349 branch of drivers for a while now. Until today drivers from that branch were only available for Windows 10 or Linux though.

It being a CUDA-wrapper is not necessarily a bad thing. It makes implementing it much easier for NVIDIA, and the real question would be how much overhead it really has. If OpenCL kernels just get compiled into CUDA kernels once, there doesn't necessarily have to be a high overhead. Unfortunately, the overhead is really hard to measure.

[NikosD]

Nvidia says the wrapper is thin, but it's not that the biggest problem of OpenCL driver.
It's not the performance.

As users of Nvidia cards write down under the anandtech's article in the comments section, there are incompatibilities, bugs and as anandtech says Nvidia didn't bother fix all those issues.
They have CUDA.

That's why they stuck years in OpenCL 1.1 version.

There was no intension of updating OpenCL driver and fixing incompatibilities and bugs.

They were completely into CUDA and they just made a wrapper to have an excuse of people complaining about complete luck of OpenCL support.

They are not lazy or incompetent.
They do it intentionally, obviously.

[huhn]

because some people in comment on this?
yes there are bugs and there even new ones in this new driver no question.

there are bugs in CUDA too and AMD has problems with openCL too.

[P.J]

So why isn't it possible to use OpenCL in some programs with Nvidia cards?

[Sulik]

OpenCL pretends to be vendor-agnostic, but it really isn't in practice... You still have to write 3 different OpenCL code paths for each vendor if you want to get any kind of decent perf. At least cuda is just like C/C++, so it's not as NVidia-centric as one might think (hence why all serious gpgpu work is done with cuda and not OpenCL).

[wanezhiling]

DXVA Checker v3.4.0 (http://bluesky23.yukishigure.com/en/DXVAChecker.html)

[huhn]

still can't use UHDcode with dxva checker...

[P.J]

DXVA Checker v3.4.0 (http://bluesky23.yukishigure.com/en/DXVAChecker.html)

Great, finally both x86-x64 together :)

[vivan]

still can't use UHDcode with dxva checker...Make sure that you:
1) Use x64 version of dxva checker
2) Installed and activated UHDcode
3) Trying to benchmark mp4 video (their splitter doesn't support mkv, and their decoder doesn't work with other splitters).
4) Checked preferred splitters/decoders...

If everything else fails you can always build a graph (and benchmark it) using Graph Studio Next (https://github.com/cplussharp/graph-studio-next) (I'm not sure if it supports DXVA, but other than that it's quite cool).

[huhn]

Make sure that you:
1) Use x64 version of dxva checker
2) Installed and activated UHDcode
3) Trying to benchmark mp4 video (their splitter doesn't support mkv, and their decoder doesn't work with other splitters).
4) Checked preferred splitters/decoders...

If everything else fails you can always build a graph (and benchmark it) using Graph Studio Next (https://github.com/cplussharp/graph-studio-next) (I'm not sure if it supports DXVA, but other than that it's quite cool).

there is no 64 bit version anymore only one version for both
the file is a 8 bit mp4
preferred splitter shouldn't matter for DXVA checker or not?

[vivan]

preferred splitter shouldn't matter for DXVA checker or not?Dunno, when I installed UHDcode I was unable to use LAV Decoder in DXVA checker until I uninstalled UHDcode (that was before NikosD said that it's enough to just switch preferred splitter).

[huhn]

graph studio next doesn't work too.

i recheck the activation but that's a very straight forward thing ....

[wanezhiling]

http://www.anandtech.com/show/9152/futureproofing-htpcs-for-the-4k-era-hdmi-hdcp-and-hevc
Damn... Skylake with Iris 6100 still have no ability for HEVC...
:devil::devil::devil:

[iSunrise]

http://www.anandtech.com/show/9152/futureproofing-htpcs-for-the-4k-era-hdmi-hdcp-and-hevc
Damn... Skylake with Iris 6100 still have no ability for HEVC...
:devil::devil::devil:
Intel Iris Graphics 6100 (as well as HD 5500, 6000) is Broadwell, not Skylake.

[wanezhiling]

Skylake-U also integrate Iris 6100.

[JohnLai]

Anyone with kepler card (especially GTX 650TI Boost) can run hybrid hevc decoding (dxva-native) without any issue?

My friend uses that kepler generation card (drivers 350.05 and 347.88) and mpc-hc 1.7.8.151 (git 3f78c2a) with the included lavfilters with EVR Custom Presenter renderer.

The problem with it, everytime when I ran any HEVC encoded video (from elecard and some files from here), I encountered some weird jerkyness.

The moment I doubleclicked the video and MPC came out, it took 25-30seconds for the video to somehow 'load'. The playback timer is still moving normally while the screen is black. After that, suddenly the video loaded and it speeds up very fast until it catched up to the 30seconds (it looks like fast forward motion). Not only that, after it able to catch up, it ran fine up for additional 10 seconds and video hang there until again for another 30 seconds, then it auto speeds up again to the duration of 1 minutes 10 seconds. These situation repeats itself in that pattern.

I checked the filters loaded by mpc-hc and only lavfilters being used.

When running any H.264 encoded video, there isn't any issue. This only happen with every HEVC files.
Can anyone with Kepler based gpu verify this?

[P.J]

Anyone with kepler card (especially GTX 650TI Boost) can run hybrid hevc decoding (dxva-native) without any issue?

My friend uses that kepler generation card (drivers 350.05 and 347.88) and mpc-hc 1.7.8.151 (git 3f78c2a) with the included lavfilters with EVR Custom Presenter renderer.

The problem with it, everytime when I ran any HEVC encoded video (from elecard and some files from here), I encountered some weird jerkyness.

The moment I doubleclicked the video and MPC came out, it took 25-30seconds for the video to somehow 'load'. The playback timer is still moving normally while the screen is black. After that, suddenly the video loaded and it speeds up very fast until it catched up to the 30seconds (it looks like fast forward motion). Not only that, after it able to catch up, it ran fine up for additional 10 seconds and video hang there until again for another 30 seconds, then it auto speeds up again to the duration of 1 minutes 10 seconds. These situation repeats itself in that pattern.

I checked the filters loaded by mpc-hc and only lavfilters being used.

When running any H.264 encoded video, there isn't any issue. This only happen with every HEVC files.
Can anyone with Kepler based gpu verify this?
Hybrid solution never worked for me even with 750Ti, that's why I sold it for GTX960

[JohnLai]

Hybrid solution never worked for me even with 750Ti, that's why I sold it for GTX960

Strange thing is that the hybrid hevc decoding works well with GTX 970 and 980.
I also can get Intel HD 4400 and 4600 hybrid HEVC decoding works well.

I have no idea why nvidia Kepler series, GTX 650TI Boost and GTX 670 have problem in hybrid decoding.

As for getting gtx 960, I dont think his finance is that great now....

[P.J]

GTX970/980 are 2nd generation of Maxwell.
Maybe, but Intel is tooooo weak...

[JohnLai]

With madvr renderer, yeah, Intel IGPU is weak, but using EVR CP, no issue at all. I just hope egur can implement QS HEVC as soon as possible.
This way, I can leave decoding to intel IGPU and leave the resizing etc for madvr renderer.

[P.J]

With madvr renderer, yeah, Intel IGPU is weak, but using EVR CP, no issue at all. I just hope egur can implement QS HEVC as soon as possible.
This way, I can leave decoding to intel IGPU and leave the resizing etc for madvr renderer.

Even with EVR/EVR CP, Intel is weak. Try to play this file:
Beauty_3840x2160_120fps_420_8bit_HEVC_MP4

[JohnLai]

PJ,
No issue with the clip when playing it with Intel HD4600 here.
Both 2 lines shown at MPC-HC EVR CP statistics are almost perfectly in sync with each other (straight lines).

Probably due to RAM. Using DDR3 2000Mhz dual channel.

[NikosD]

OpenCL 1.2 has been available in the 349 branch of drivers for a while now. Until today drivers from that branch were only available for Windows 10 or Linux though.


My Fermi card (GT610) under Windows 10 build 10049 has no option of downloading any Microsoft Update driver.

So, I was forced to use a Win 8.1 driver (347.25), but unfortunately under Win 10 has no OpenCL at all.
Not even OpenCL 1.1.

Weird decision of Nvidia for Fermi owners.

I'll wait for the next builds of Win 10 to check for an update.

[nevcairiel]

Fermi does not support OpenCL 1.2, only Kepler and Maxwell.

[NikosD]

I know, Nvidia decided to give DX12 to Fermi but not OpenCL 1.2
As I wrote above, the problem is that under Win 10 Fermi doesn't even support OpenCL in general.

[P.J]

Someone please test PowerDVD 15 :)

[ibius]

And also Lentoid 2.0.3.1, which now has x64 version.

I did a quick test on my Intel DualCore with SSE4.1 and my brother's AMD X4 with SSE4A, and compared Lentoid x64 vs LAV x64 (nightly), tested two common 4k24p clips.
On my DualCore the performance of Lentoid was about the same as LAV's, but on my brother's QuadCore Lentoid beat LAV badly, I guess LAV's 64-bit optimizations don't include SSE4A.

[NikosD]

I made a benchmark test using latest versions of various decoders and tools on 3 different CPU architecture systems using HEVC video files of 1080p and 2160p

Decoders:
LAV x64/x86 0.65
MPC Video Decoder x64/x86 1.4.4.286
Lentoid x64/x86 2.0.3.2
Cyberlink HEVC PowerDVD 15.0.1510.58
Microsoft HEVC decoder build 10049

Tools:
DXVA Checker 3.4.1 in Benchmark Decode mode

Splitter:
LAV Splitter x64/x86 0.65


System 1 (16 threads for MPC)

Core i7-4790(AVX2) - Win 8.1 Pro x64



1.ProRes-1080p@30fps-2Mbps



MPC x64 303/421/495 CPU 85%

Lentoid x64 301/407/467 CPU 87%

Lentoid x86 315/384/450 CPU 88%

LAV x64 275/380/441 CPU 81%

MPC x86 145/185/252 CPU 93%

LAV x86 136/176/263 CPU 92%

PDVD x86 95/164/202 CPU 28%




2.Elephants Dream-1080p@24fps-1.7Mbps



MPC x64 135/372/576 CPU 62%

Lentoid x64 138/362/605 CPU 65%

LAV x64 135/362/605 CPU 55%

Lentoid x86 118/310/480 CPU 68%

MPC x86 75/171/420 CPU 67%

PDVD x86 65/169/259 CPU 24%

LAV x86 70/156/380 CPU 68%



3.Big Buck Bunny-1080p@60fps-2Mbps



LAV x64 265/419/605 CPU 53%

MPC x64 263/418/571 CPU 56%

Lentoid x64 185/405/570 CPU 59%

Lentoid x86 171/363/522 CPU 62%

PDVD x86 134/198/259 CPU 22%

MPC x86 83/182/394 CPU 56%

LAV x86 72/172/410 CPU 53%



4. Girls -1080p60 fps -11Mbps



LAV x64 214/297/378 CPU 68%

Lentoid x64 2220/296/365 CPU 84%

Lentoid x86 214/289/361 CPU 85%

MPC x64 204/283/360 CPU 71%

PDVD x86 120/172/220 CPU 59%

MPC x86 106/152/202 CPU 86%

LAV x86 100/144/206 CPU 84%



5.Beauty-2160p@30fps-12.3Mbps



MPC x64 83/94/98 CPU 77%

LAV x64 78/90/95 CPU 71%

Lentoid x64 68/83/92 CPU 76%

Lentoid x86 68/83/88 CPU 79%

MPC x86 34/50/56 CPU 91%

LAV x86 35/47/49 CPU 93%

PDVD x86 31/37/41 CPU 29%



6.Fitness-2160p@30fps-8Mbps



LAV x64 99/113/138 CPU 84%

MPC x64 90/112/130 CPU 92%

Lentoid x64 67/102/119 CPU 92%

Lentoid x86 63/95/128 CPU 93%

PDVD x86 70/77/97 CPU 91%

MPC x86 42/63/110 CPU 96%

LAV x86 37/61/113 CPU 94%



7.Ducks-2160p@50fps-4Mbps



LAV x64 107/119/121 CPU 91%

MPC x64 102/116/119 CPU 94%

Lentoid x64 88/106/111 CPU 91%

Lentoid x86 74/94/99 CPU 91%

PDVD x86 77/83/87 CPU 78%

MPC x86 40/50/53 CPU 91%

LAV x86 37/48/50 CPU 94%



8.Gravity - 2048x858@24fps-5Mbps-10bit



LAV x64 125/202/404 CPU 60%

MPC x64 123/201/396 CPU 63%

MPC x86 80/130/360 CPU 67%

LAV x86 76/126/379 CPU 65%

PDVD x86 75/117/265 CPU 53%

Lentoid x64/x86 Don't support 10bit HEVC



System 2 (8 threads for MPC)

Core 2 Quad@2.26GHz(SSE4.1) - Win 10 x64 build 10049



1.ProRes-1080p@30fps-2Mbps



Lentoid x64 52/96/113 85%

LAV x64 82/95/102 CPU 89%

MPC x64 75/95/102 CPU 89%

Lentoid x86 48/93/110 86%

Microsoft x64 65/77/84 CPU 88%

Microsoft x86 57/67/72 CPU 87%

PDVD x86 34/63/75 CPU 63%

MPC x86 45/60/81 CPU 95%

LAV x86 41/57/80 CPU 93%




2.Elephants Dream-1080p@24fps-1.7Mbps



LAV x64 43/90/108 CPU 80%

MPC x64 49/87/102 CPU 87%

Lentoid x64 44/84/96 CPU 85%

Lentoid x86 43/84/98 CPU 85%

Microsoft x64 49/70/79 CPU 87%

Microsoft x86 42/66/73 CPU 85%

PDVD x86 25/61/79 CPU 59%

LAV x86 25/54/107 CPU 83%

MPC x86 29/56/87 CPU 89%



3.Big Buck Bunny-1080p@60fps-2Mbps



LAV x64 56/92/102 CPU 80%

Lentoid x64 58/90/99 CPU 82%

MPC x64 64/90/98 CPU 87%

Lentoid x86 57/89/98 CPU 83%

Microsoft x64 66/78/86 CPU 85%

Microsoft x86 56/69/76 CPU 83%

PDVD x86 40/68/80 CPU 60%

MPC x86 30/63/103 CPU 86%

LAV x86 27/58/102 CPU 79%



4. Girls -1080p60 fps -11Mbps



LAV x64 62/86/107 CPU 87%

MPC x64 62/80/93 CPU 91%

Lentoid x64 53/80/94 CPU 90%

Lentoid x86 55/79/93 CPU 91%

Microsoft x64 54/68/77 CPU 88%

Microsoft x86 38/54/73 CPU 85%

PDVD x86 38/54/73 CPU 76%

MPC x86 33/50/84 CPU 93%

LAV x86 32/49/86 CPU 92%



5.Beauty-2160p@30fps-12.3Mbps



LAV x64 13/21/24 CPU 83%

MPC x64 11/20/22 CPU 90%

Lentoid x64 5/20/23 CPU 84%

Lentoid x86 4/19/23 CPU 83%

Microsoft x64 7/16/18 CPU 90%

MPC x86 5/16/18 CPU 96%

LAV x86 6/15/18 CPU 94%

Microsoft x86 6/14/18 CPU 88%

PDVD x86 8/13/16 CPU 64%



6.Fitness-2160p@30fps-8Mbps



LAV x64 20/25/27 CPU 92%

MPC x64 16/22/24 CPU 93%

Lentoid x64 4/21/22 92%

Lentoid x86 4/21/23 93%

MPC x86 11/19/26 CPU 95%

LAV x86 9/19/28 CPU 95%

PDVD x86 15/17/20 CPU 89%

Microsoft x64 11/17/20 CPU 89%

Microsoft x86 8/16/18 CPU 86%



7.Ducks-2160p@50fps-4Mbps



LAV x64 18/24/25 CPU 94%

MPC x64 14/22/23 CPU 92%

Lentoid x64 5/21/24 CPU 91%

Lentoid x86 5/21/22 CPU 93%

PDVD x86 13/17/17 CPU 86%

Microsoft x64 9/17/18 CPU 90%

Microsoft x86 7/15/16 CPU 86%

LAV x86 9/15/17 CPU 95%

MPC x86 7/15/18 CPU 96%



8.Gravity - 2048x858@24fps-5Mbps-10bit



LAV x64 36/50/70 CPU 89%

MPC x64 36/49/69 CPU 90%

MPC x86 29/44/71 CPU 92%

LAV x86 28/44/72 CPU 90%

Microsoft x64 29/39/49 CPU 84%

PDVD x86 28/39/49 CPU 71%

Microsoft x86 23/34/53 CPU 84%

Lentoid x64/x86 Don't support 10bit HEVC




System 3 (4 threads for MPC)

Core 2 Duo@2.33GHz (SSSE3) - Win 8 Pro x64



1.ProRes-1080p@30fps-2Mbps



Lentoid x64 40/64/94 CPU 90%

Lentoid x86 39/61/87 CPU 89%

PDVD x86 31/52/73 CPU 84%

MPC x64 26/37/65 CPU 95%

LAV x64 23/35/62 CPU 92%

MPC x86 22/32/60 CPU 94%

LAV x86 20/30/66 CPU 92%



2.Elephants Dream-1080p@24fps-1.7Mbps



Lentoid x64 22/66/120 CPU 90%

Lentoid x86 22/64/116 CPU 90%

PDVD x86 19/56/102 CPU 81%

MPC x64 16/38/97 CPU 93%

LAV x64 15/35/83 CPU 90%

MPC x86 16/34/93 CPU 94%

LAV x86 13/31/90 CPU 89%



3.Big Buck Bunny-1080p@60fps-2Mbps



Lentoid x64 36/83/123 CPU 89%

Lentoid x86 35/79/118 CPU 89%

PDVD x86 31/68/104 CPU 80%

MPC x64 20/44/98 CPU 92%

LAV x64 16/39/85 CPU 89%

MPC x86 17/38/94 CPU 92%

LAV x86 13/36/91 CPU 88%



4. Girls -1080p60 fps -11Mbps



Lentoid x64 29/48/83 CPU 89%

Lentoid x86 28/47/76 CPU 91%

PDVD x86 25/40/73 CPU 84%

MPC x64 19/31/68 CPU 93%

LAV x64 17/29/65 CPU 91%

MPC x86 16/27/73 CPU 93%

LAV x86 14/25/60 CPU 91%



5.Beauty-2160p@30fps-12.3Mbps



Lentoid x64 4/14/16 CPU 90%

Lentoid x86 3/14/16 CPU 91%

PDVD x86 7/12/15 CPU 78%

MPC x64 2/10/12 CPU 95%

LAV x64 1/9/10 CPU 91%

MPC x86 1/8/10 CPU 93%

LAV x86 1/8/9 CPU 92%



6.Fitness-2160p@30fps-8Mbps



Lentoid x64 4/17/27 CPU 88%

PDVD x86 11/16/24 CPU 92%

Lentoid x86 4/16/27 CPU 87%

MPC x64 5/13/23 CPU 93%

LAV x64 5/12/23 CPU 92%

MPC x86 4/11/22 CPU 92%

LAV x86 4/10/22 CPU 92%



7.Ducks-2160p@50fps-4Mbps



Lentoid x64 7/16/18 CPU 87%

Lentoid x86 6/16/18 CPU 87%

PDVD x86 11/15/17 CPU 88%

MPC x64 4/9/11 CPU 92%

LAV x64 3/9/11 CPU 90%

MPC x86 2/8/10 CPU 92%

LAV x86 2/8/9 CPU 91%



8.Gravity - 2048x858@24fps-5Mbps-10bit



MPC x64 18/33/85 CPU 94%

LAV x64 17/32/84 CPU 91%

PDVD x86 18/31/79 CPU 82%

MPC x86 15/28/83 CPU 93%

LAV x86 13/27/84 CPU 91%

Lentoid x64/x86 Don't support 10bit HEVC




Comments:


1) Lentoid x86 is an unbelievably fast decoder reaching close to all x64 decoders even in newer architectures (SSE4.1)
Lentoid x86 is the fastest x86 HEVC decoder for all OSes and all CPU architectures.

On the other hand, LAV x86 is always the slowest (besides 10bit clips on newer architectures)

2) Lentoid x64 is fast and occasionally faster in one or two clips for 1080p, but for all the other cases is a little slower than LAV/MPC x64 (besides pre-SSE4.1 which is always faster)

3) Lentoid still doesn't support 10 bit HEVC clips.

4) LAV x64 is fast, most of the times faster than any other HEVC decoder, if and only if, CPU architecture uses at least SSE4.1 instructions. i.e Penryn (2008) and above for Intel and Bulldozer (late 2011) for AMD.

For older systems e.g Core2Duo with SSSE3 and older, Lentoid x64/x86 and PowerDVD 15 are faster, especially Lentoid.

Even Lentoid x86 is two times faster than LAV x64 in pre-SSE4.1 systems

5) Microsoft added MKV support for HEVC files, but Microsoft's HEVC decoders don't look so fast under SSE4.1 CPU architecture like they did under SSSE3 CPU architecture, especially now that Lentoid has built so fast x64/x86 decoders.
PowerDVD 15 is also competitive with Microsoft's decoders in 2160p videos.

6) I couldn't figure out what is going on with the CPU utilization of PowerDVD 15 decoder.
At the first three mkv files, I thought that it's a 2 thread decoder, but in the forth mp4 file the CPU usage went higher, but not in the level of the other decoders.
It looks like it has room for optimizations.

PowerDVD 15 looks very strong for old architectures (before SSE4.1), especially for 2160p

7) MPC decoder is based on LAV decoder and the results are similar.

8) HEVC 2160p is unreachable for older Core 2 Quad cores with low clock, even using SSE4.1 and low bitrates.

9) Lentoid x64/x86 uses YV12 output during playback and I420 output in decode mode.
PowerDVD 15 uses YUY2 for playback/decode for both 8bit/10bit clips and LAV/MPC use NV12 in both playback and decode mode for 8bit but for 10bit clips use P010 for decode and NV12 for playback.

[iSunrise]

Holy shit, LAV x64 is about two times as fast than LAV x86 with HEVC and even ProRes. Thank god madVR is finally available as a x64 build.

[sneaker_ger]

I think "ProRes" is just the name of the sample. The original source was probably a ProRes file before being encoded to HEVC.

[NikosD]

I think "ProRes" is just the name of the sample. The original source was probably a ProRes file before being encoded to HEVC.
Exactly

[P.J]

Ah, PowerDVD 15 doesn't support 10bit...

[ibius]

Are you sure? PowerDVD 14, or at least it's CyberLink Video Decoder played 10bit HEVC clips just fine in MPC-HC and DXVAChecker.

Also, someone with a Kepler/Maxwell card and new Nvidia drivers should try CyberLink's hybrid HEVC decoding again, now that Nvidia "supports" OpenCL 1.2, because in SW-mode it most likely won't beat LAV's or Lentoid's x64 versions.

[P.J]

I meant 10-bit via DXVA2, 8-bit works fine now. No way to benchmark it with DXVAChecker too even after registering the decoder =T

[huhn]

Are you sure? PowerDVD 14, or at least it's CyberLink Video Decoder played 10bit HEVC clips just fine in MPC-HC and DXVAChecker.

Also, someone with a Kepler/Maxwell card and new Nvidia drivers should try CyberLink's hybrid HEVC decoding again, now that Nvidia "supports" OpenCL 1.2, because in SW-mode it most likely won't beat LAV's or Lentoid's x64 versions.

nvidia user are now able to use openCL 1.2 that doesn't mean it is faster at all. it simply means they could write a openCL program with some of the new feature from openCL 1.2 that are in the new driver and maybe this give more speed. but a new version of this decoder is needed first. and i highly doubt this give more speed.

[NikosD]

Ibius put the issue of OpenCL PowerDVD HEVC decoder regarding Nvidia, not in terms of performance but in terms of compatibility.

Because when something doesn't work at all, like OpenCL HEVC decoder for Nvidia, you don't care about speed.

All you want is a working decoder.

And because Cyberlink released a new OpenCL HEVC decoder v15 and Nvidia released a new driver supporting a newer version of OpenCL for Kepler/Maxwell, one might think that at last OpenCL HEVC decoder is working for Nvidia.
(I doubt that there are devs interested in Nvidia's OpenCL, but who knows)

[ibius]

I meant 10-bit via DXVA2, 8-bit works fine now.

I see, and I meant SW-decoding, because CyberLink's Video Decoder from PowerDVD 14 didn't even have 8-bit HEVC DXVA2 decoding, only hybrid decoding (OpenCL) and only on AMD and Intel cards.


Ibius put the issue of OpenCL PowerDVD HEVC decoder regarding Nvidia, not in terms of performance but in terms of compatibility.

Because when something doesn't work at all, like OpenCL HEVC decoder for Nvidia, you don't care about speed.

All you want is a working decoder.

Exactly, thank you.

Why both CyberLink's and Strongene's OpenCL HEVC decoders work on AMD and Intel cards, but not on NVIDIA?

[huhn]

Why both CyberLink's and Strongene's OpenCL HEVC decoders work on AMD and Intel cards, but not on NVIDIA?

they all use different APIs. looks like amd and intel have the similar APIs or the same that are needed for this program while nvidia has it own to get the same results.

[ibius]

they all use different APIs. looks like amd and intel have the similar APIs or the same that are needed for this program while nvidia has it own to get the same results.

It was a rhetorical question. Two independent decoders that utilize OpenCL work on AMD and Intel, but not on NVIDIA.

Like I said before at least for hybrid decoding purposes NVIDIA's OpenCL is a joke, although I'm pretty happy with my GTX 550Ti madVR-wise, I'm gonna have to upgrade to GTX 960 or similar when HEVC encodes hit main street.

[huhn]

It was a rhetorical question. Two independent decoders that utilize OpenCL work on AMD and Intel, but not on NVIDIA.

Like I said before at least for hybrid decoding purposes NVIDIA's OpenCL is a joke, although I'm pretty happy with my GTX 550Ti madVR-wise, I'm gonna have to upgrade to GTX 960 or similar when HEVC encodes hit main street.
these decoder are decoder for AMD and intel and simply not nvidia so they are not independent at all.

nvidia DXVA for HEVC using DXVA (except 960) is hybrid like these other openCL decoder. compared to lavfilter x64 or the windows 10 decoder all hybrid decoder are looking like a joke. maybe faster but at which cost?

and nvidia is usually no challenge for AMD in term of GPGPU processing. no secret and cuda doesn't change that.

[ibius]

these decoder are decoder for AMD and intel and simply not nvidia so they are not independent at all.

That's a nice phrase "for AMD and Intel and simply not Nvidia", so Strongene and CyberLink developed their OpenCL HEVC decoders specifically for AMD and Intel, but not Nvidia?
And how come it follows that they are not idependent?

In Strongene's OpenCL HEVC decoder readme file there's a note that says The OpenCL Version Lentoid HEVC decoder now only guarantees the support for AMD dGPUs and APUs (Radeon HD 5000 series or above).
Which implies that an earlier version wasn't AMD-specific.

And I didn't see any AMD-only developer notes from CyberLink, so how do you know that both Strongene and CyberLink only implemented AMD/Intel-specific APIs?
If two different decoders with OpenCL support only work with AMD and Intel cards doesn't mean both decoders were developed only for AMD and Intel cards.
Maybe if Nvidia gets to OpenCL 2.0 in a few years, these then old decoders will suddenly start working.

nvidia DXVA for HEVC using DXVA (except 960) is hybrid like these other openCL decoder. compared to lavfilter x64 or the windows 10 decoder all hybrid decoder are looking like a joke.

That's not what I meant or said at all, please pay more attention next time.

[huhn]

if they wanted to write a openCL decoder for nvidia than they could have done that. but they didn't. why? feel free to ask them.
but it is possible! and openCL 1.2 or 2.0 is for sure not the reason.

[NikosD]

You have to be a little or more masochist to write a program in OpenCL for Nvidia when there is CUDA, that's the truth.

So, you write an OpenCL program for the HW that truly supports it (AMD, Intel) and if you have even more resources available, you write another version for CUDA.

End of story.

[NikosD]

A very interesting update:

PowerDVD 15 seems to support 3 HEVC decoding modes, at least for Intel iGPU (Haswell).

CPU, DXVA and OpenCL.

Expect an updated post testing all these modes.

[huhn]

You have to be a little or more masochist to write a program in OpenCL for Nvidia when there is CUDA, that's the truth.

So, you write an OpenCL program for the HW that truly supports it (AMD, Intel) and if you have even more resources available, you write another version for CUDA.

End of story.

i hope you are aware that there are a lot of programs that work totally fine on all 3 GPU vendor. and i hope these people ignore such commentss and still give nvidia users the chance of using there openCL programs.

[ibius]

And I hope you are aware that this is a thread for HEVC decoder evaluation, and the only two OpenCL decoders available don't work on Nvidia hardware, period.

The only HW-assisted HEVC decoding for Nvidia users is through DXVA2 (w/ LAV and maybe PowerDVD 15 now) on high-end Kepler and Maxwell cards or GTX 960 with dedicated ASIC.

[huhn]

And I hope you are aware that this is a thread for HEVC decoder evaluation, and the only two OpenCL decoders available don't work on Nvidia hardware, period.

The only HW-assisted HEVC decoding for Nvidia users is through DXVA2 (w/ LAV and maybe PowerDVD 15 now) on high-end Kepler and Maxwell cards or GTX 960 with dedicated ASIC.

and i wonder why?

The OpenCL version is developed jointly by Strongene and AMD

[huhn]

mircosoft is faster on this sample than lavfilter on my system.

i find this pretty strange and i don't see a reason for this.
and i got UHDcode working and it is indeed hybrid decoding. but with no real impact

i3 4130 r9 270
beauty
playback 1280x720

microsoft
59 FPS 89% CPU 26% GPU
UHDcode
34.5 FPS 65% CPU 51% GPU
LAV
46.5 FPS CPU 73% ~8GPU

DECODE
microsoft
59.5 FPS 89% CPU - GPU
UHDcode
36.5 FPS 68% CPU 30 % GPU
LAV
49 FPS 73% CPU - GPU

is it new that microsoft MF can read HEVC in MKV?

[NikosD]

I've finished my little survey today about HEVC CPU decoders, updated with new decoders and new tools here:
http://forum.doom9.org/showthread.php?p=1717938#post1717938

I have to admit that I was strongly impressed by Lentoid x86 decoder, more than anything else.

So, I searched a little bit in the past in order to see some signs of this incredible performance.

I had already tested the previous version 2.0.2.6 twice, first time in the initial post and second time here:
http://forum.doom9.org/showthread.php?p=1705352#post1705352

The results weren't that impressive and now I can tell that there are two reasons behind that.

In order to see for yourselves the two reasons I tested again head to head the two Lentoid versions 2.0.2.6 vs 2.0.3.x in all three systems on two files - 4 and 5 - for both modes playback (using scaling to 1280x720) and decode:

* for Lentoid 2.0.3.3 I used latest DXVA Checker v3.50 (not the older 3.4.1)

System 1

Core i7-4790(AVX2) - Win 8.1 Pro x64 - iGPU@1.5GHz - v4156


4. Girls -1080p60 fps -11Mbps

Lentoid x86 v2.0.3.3(decode) 208/276/332 CPU 84%
Lentoid x86 v2.0.3.2(decode) 214/289/361 CPU 85%
Lentoid x86 v2.0.3.1(decode) 213/288/358 CPU 86%
Lentoid x86 v2.0.2.6(decode) 94/154/210 CPU 35%


Lentoid x86 v2.0.3.3(playback) 89/107/109 CPU 21%
Lentoid x86 v2.0.3.2(playback) 89/107/108 CPU 21%
Lentoid x86 v2.0.3.1(playback) 88/108/109 CPU 20%
Lentoid x86 v2.0.2.6(playback) 74/100/109 CPU 22%


5.Beauty-2160p@30fps-12.3Mbps


Lentoid x86 v2.0.3.3(decode) 65/81/86 CPU 79%
Lentoid x86 v2.0.3.2(decode) 68/83/88 CPU 79%
Lentoid x86 v2.0.3.1(decode) 68/83/89 CPU 80%
Lentoid x86 v2.0.2.6(decode) 35/45/48 CPU 36%


Lentoid x86 v2.0.3.3(playback) 28/48/49 CPU 37%
Lentoid x86 v2.0.3.2(playback) 27/47/47 CPU 35%
Lentoid x86 v2.0.3.1(playback) 25/47/50 CPU 36%
Lentoid x86 v2.0.2.6(playback) 15/33/35 CPU 24%



System 2

Core 2 Quad@2.26GHz(SSE4.1) - Win 10 x64 build 10049 - GT610 - 347.25


4. Girls -1080p60 fps -11Mbps

Lentoid x86 v2.0.3.2(decode) 55/79/93 CPU 91%
Lentoid x86 v2.0.3.1(decode) 56/75/92 CPU 90%
Lentoid x86 v2.0.2.6(decode) 38/68/94 CPU 72%


Lentoid x86 v2.0.3.2(playback) 29/44/45 CPU 43%
Lentoid x86 v2.0.3.1(playback) 32/44/44 CPU 45%
Lentoid x86 v2.0.2.6(playback) 32/43/45 CPU 44%


5.Beauty-2160p@30fps-12.3Mbps

Lentoid x86 v2.0.3.2(decode) 4/19/23 CPU 83%
Lentoid x86 v2.0.3.1(decode) 4/20/23 CPU 82%
Lentoid x86 v2.0.2.6(decode) 8/18/21 CPU 71%


Lentoid x86 v2.0.3.2(playback) 1/13/13 CPU 40%
Lentoid x86 v2.0.3.1(playback) 1/13/15 CPU 41%
Lentoid x86 v2.0.2.6(playback) 3/12/13 CPU 39%



System 3

Core 2 Duo@2.33GHz (SSSE3) - Win 8 Pro x64 - iGPU 945GM - 8.15.10.1930



4. Girls -1080p60 fps -11Mbps


Lentoid x86 v2.0.3.3(decode) 26/46/90 CPU 89%
Lentoid x86 v2.0.3.2(decode) 28/47/76 CPU 91%
Lentoid x86 v2.0.3.1(decode) 28/46/75 CPU 90%
Lentoid x86 v2.0.2.6(decode) 24/41/81 CPU 88%


Lentoid x86 v2.0.3.3(playback) 27/38/45 CPU 72%
Lentoid x86 v2.0.3.2(playback) 26/37/45 CPU 72%
Lentoid x86 v2.0.3.1(playback) 22/37/45 CPU 73%
Lentoid x86 v2.0.2.6(playback) 20/30/44 CPU 63%


5.Beauty-2160p@30fps-12.3Mbps


Lentoid x86 v2.0.3.3(decode) 5/15/17 CPU 92%
Lentoid x86 v2.0.3.2(decode) 3/14/16 CPU 91%
Lentoid x86 v2.0.3.1(decode) 3/14/16 CPU 93%
Lentoid x86 v2.0.2.6(decode) 4/12/14 CPU 84%


Playback mode using EVR not supported by iGPU for UHD.


A careful reader will see the two points that made new Lentoid x86 2.0.3.x extremely fast:

1) In the first two benchmarks tests, I used DXVA processing or Playback benchmark mode as it is called now.

Well, I don't know if it's a problem of DXVA Checker or of Lentoid x64/x86 decoder, but in this mode the CPU utilization of Lentoid is low and the performance is low too.
It's like a bottleneck across all systems and both versions old (2.0.2.6) and new (2.0.3.x)

I've sent an email to the DXVA Checker's developer about this issue, but he hasn't confirmed anything yet.

In the last test, I used Decode Benchmark mode, where the Lentoid x64/x86 decoders really shine, with no bottlenecks and high CPU utilization.

2) In my comments of second test on January 2015 I wrote:


Lentoid (Strongene) seems to be the fastest x86 decoder, but not with HT CPUs. It looks like a problem with multi-threading.


Well, it seems that Lentoid has already seen that weakness of their decoder (or they read my benchmarks :p) and they fixed exactly that.

See for yourself the double speed of Lentoid x86 in Decode mode for System 1 (Core i7 4790 - 8 threads) because of the double CPU utilization of the new version compared to the old one on the same system.

Also for the first system, there is an improvement in Playback Benchmark mode, compared to the older platforms, but only for the 2160p file and it's still too low performance for such a fast decoder.

I hope a few things are better clarified now.

[NikosD]

is it new that microsoft MF can read HEVC in MKV?

From my 8 HEVC samples, the 5 of them are .mkv and they were decoded by Microsoft x64/x86 HEVC decoder using LAV splitter and Win 10 Build 10049 - System 2

[ksio89]

I'm trying to play some video samples encoded in HEVC but DXVA2 is not being activated, does anyone know why? Both HEVC and UHD are available and activated in LAV (both x86 and x64) when DXVA2 (native) is chosen as the HW decoder. BTW, I'm using MPC-HC x64.

Some time ago I posted a similar question, but it was an UHD video in H.264 codec instead of HEVC.

The video samples are available at http://www.elecard.com/en/download/videos.html

Tried HEVC in UHD, FHD, and HD, DXVA2 isn't activated with none of them.


These are the specs of my laptop:

Dell Inspiron 15R-5537

CPU: core i7-4500U @ 1,8-2,7 GHz
GPU: AMD Radeon HD 8870M
RAM: 8 GB DDR3 @ 1600 MHz
OS: Win 8.1 Pro

software info:

codec: K-Lite codec pack 11.1.0 Full
MPC-HC build: 1.7.8.152 (3f78c2a) (develop) x64
LAV filters: 0.64.0.51

This laptop is an Enduro (MUXless dynamic) platform, similar to nVidia Optimus, in which the iGPU is activated all the time, but rendering (gaming, video playback) can be done by GPU via PCI-E bus if set to high perfomance. However, as someone pointed before, the Radeon GPU can only decode HEVC via OpenCL, so DXVA HW decoding of HEVC is only available to HD Graphics 4400 iGPU.

I don't know if HD Graphics 4400 HEVC decoder or the CPU is fast enough to play 4K, but DXVA should be at least activated, because HEVC is shown as available when DXVA2 native/copy-back HW decoder is selected. Thanks in advance.

[NikosD]

Can you post a DXVA checker first screen screenshot ?

I think your Radeon card "overlaps" your iGPU so only DXVA capabilities of Radeon would be available.

If that's true, you have to wait for egur, Intel's employee, to add to his QS decoder HEVC decoding in order to use it with your laptop.

[NikosD]

BTW, I'm using MPC-HC x64.

Some time ago I posted a similar question, but it was an UHD video in H.264 codec instead of HEVC.



If you go to LAV Video properties inside MPC-HC and choose QS decoder (Intel QuickSync) and you tick UHD (4K), you will have HW acceleration for H.264 UHD (4K) files.

[littleD]

Did anybody notice that Strongene Lentoid decoder has also 64 bit version now? And not suprisingly, it is a bit faster than x86 version. Just tiny behind Lav 64 bit.

[wanezhiling]

Did anybody notice that Strongene Lentoid decoder has also 64 bit version now? And not suprisingly, it is a bit faster than x86 version. Just tiny behind Lav 64 bit.

http://forum.doom9.org/showthread.php?p=1717847#post1717847

[NikosD]

This is a post for hybrid decoding (DXVA, OpenCL) on Intel's iGPU HD4600.


Decoders:
LAV x64/x86 DXVA 0.65
MPC x64/x86 DXVA 1.4.4.286
PowerDVD x86 DXVA & OpenCL 15.0.1510.58
Lentoid x86 OpenCL decoder v2.0.0.2



Tool:
DXVA Checker 3.4.1 (Benchmark Decode mode)


Splitter:
LAV Splitter 0.65


All DXVA tests were done in Native mode.
GPU utilization is ~100% in almost all tests during all the time, with the exceptions of LAV/MPC written in comments.


System 1

Core i7-4790 - Win 8.1 Pro x64 - iGPU HD4600@1.5GHz - v4156


1.ProRes-1080p@30fps-2Mbps




MPC DXVA x64 135/203/229 CPU 21%

LAV DXVA x64 131/202/224 CPU 22%

LAV DXVA x86 135/202/227 CPU 24%

PDVD DXVA x86 137/200/237 CPU 25%

PDVD OCL x86 142/198/222 CPU 34%

MPC DXVA x86 145/197/225 CPU 23%

Lentoid OCL x86 (SW fall-back mode)



2.Elephants Dream-1080p@24fps-1.7Mbps



PDVD DXVA x86 132/219/308 CPU 28%

LAV DXVA x64 87/211/304 CPU 23%

MPC DXVA x64 94/208/295 CPU 22%

PDVD OCL x86 120/203/247 CPU 35%

LAV DXVA x86 74/203/290 CPU 25%

MPC DXVA x86 75/202/293 CPU 25%

Lentoid OCL x86 Black screen



3.Big Buck Bunny-1080p@60fps-2Mbps



LAV DXVA x64 180/255/312 CPU 23%

MPC DXVA x64 165/253/308 CPU 27%

PDVD DXVA x86 196/249/300 CPU 28%

MPC DXVA x86 152/242/298 CPU 25%

LAV DXVA x86 135/241/303 CPU 25%

PDVD OCL x86 172/218/241 CPU 33%

Lentoid OCL x86 Not working



4. Girls -1080p60 fps -11Mbps



PDVD DXVA x86 160/199/242 CPU 41%

PDVD OCL x86 140/181/207 CPU 46%

MPC DXVA x64 109/170/218 CPU 27%

LAV DXVA x64 144/166/215 CPU 28%

MPC DXVA x86 91/157/213 CPU 27%

LAV DXVA x86 106/153/196 CPU 27%

Lentoid OCL x86 (SW fall-back mode)



5.Beauty-2160p@30fps-12.3Mbps



LAV DXVA x64 44/54/56 CPU 22%

PDVD DXVA x86 48/53/58 CPU 25%

MPC DXVA x64 18/51/56 CPU 21%

LAV DXVA x86 46/49/52 CPU 25%

MPC DXVA x86 Crashed in 3rd time in a row test

Lentoid OCL x86 (SW fall-back mode)

PDVD OCL x86 (SW fall-back mode)



6.Fitness-2160p@30fps-8Mbps



PDVD DXVA x86 66/75/92 CPU 57%

MPC DXVA x64 41/60/84 CPU 33%

LAV DXVA x64 49/58/82 CPU 34%

LAV DXVA x86 47/57/82 CPU 33%

MPC DXVA x86 37/56/82 CPU 33%

PDVD OCL x86 (SW fall-back mode)

Lentoid OCL x86 Crashed



7.Ducks-2160p@50fps-4Mbps



PDVD DXVA x86 62/67/70 CPU 26%

LAV DXVA x64 63/66/68 CPU 26%

LAV DXVA x86 63/66/68 CPU 27%

MPC DXVA x64 35/63/69 CPU 24%

MPC DXVA x86 Crashed in 3rd time in a row test

Lentoid OCL x86 (SW fall-back mode)

PDVD OCL x86 (SW fall-back mode)




8.Gravity - 2048x858@24fps-5Mbps-10bit

Intel's hybrid decoder doesn't support 10 bit HEVC decoding, so all decoders work but in SW fall-back mode with the exception of Lentoid OCL which crashed.



Comments:



0) 4K60fps decoding is feasible using DXVA hybrid decoders, only with low bitrates on HD4600@1.5GHz.

1) PowerDVD 15 DXVA x86 decoder is not only the fastest x86 DXVA decoder for HEVC, but also the fastest DXVA decoder in general, as it is faster, most of the times, even than LAV/MPC x64 HEVC decoders.

2) On the other hand, PowerDVD 15 OCL x86 decoder although is fast, it uses always more CPU power than DXVA hybrid decoders and mainly it doesn't support 2160p decoding and falls back to SW.

3) LAV/ MPC x64 versions are a little faster than the x86 versions, but not with a big difference.

LAV/MPC x64/x86 had some problems utilizing the GPU completely, sometimes they dropped the load a lot and iGPU frequency dropped too, resulting in lower performance e.g clips 2,4,6.

Optimizations needed for LAV DXVA in order to be competitive with PowerDVD 15 DXVA utilizing GPU at 100% all the time.

4) Lentoid OpenCL is simply not working. Major compatibilities with a lot of HEVC files and when it's working it falls back to SW mode.

5) All DXVA decoders output is NV12, PowerDVD 15 OCL uses YUY2 and Lentoid OCL uses I420

[P.J]

How did you add PDVD to DXVAChecker?

[NikosD]

What do you mean ?

The registration of the DS filter and the use of it by DXVA Checker is exactly the same as PowerDVD 14.

[nevcairiel]

If PowerDVD outputs YUY2, then it should not be used. YUY2 is a 4:2:2 format, which means it upsamples the chroma - you do not get the original image from the decoder!

[huhn]

3) LAV/ MPC x64 versions are a little faster than the x86 versions, but not with a big difference.

LAV/MPC x64/x86 had some problems utilizing the GPU completely, sometimes they dropped the load a lot and iGPU frequency dropped too, resulting in lower performance e.g clips 2,4,6.

Optimizations needed for LAV DXVA in order to be competitive with PowerDVD 15 DXVA utilizing GPU at 100% all the time.

because the DXVA decoder is in the driver it self it should come down to the buffering and not sure if this should be judge in benchmark mode.

[NikosD]

If PowerDVD outputs YUY2, then it should not be used. YUY2 is a 4:2:2 format, which means it upsamples the chroma - you do not get the original image from the decoder!

Yes, PowerDVD OCL prefers YUY2 at least in decode mode and IIRC in Playback too.

Of course PowerDVD DXVA uses the mandatory NV12.

because the DXVA decoder is in the driver it self it should come down to the buffering and not sure if this should be judge in benchmark mode.

I wrote above that PowerDVD DXVA using the exact same DXVA decoder device on the same clip, it pushes the GPU load to ~100% all the time.

[P.J]

What do you mean ?

The registration of the DS filter and the use of it by DXVA Checker is exactly the same as PowerDVD 14.

What file exactly? It doesn't work for me at all :confused:

http://phota.me/0d4y.png

[NikosD]

Look an older post of Ibius here in previous pages, around January with detailed directions about PowerDVD DSfilter use.

[NikosD]

If PowerDVD outputs YUY2, then it should not be used. YUY2 is a 4:2:2 format, which means it upsamples the chroma - you do not get the original image from the decoder!

I just saw something interesting.

PDVD OCL implementation uses the DXVA Decoder Device of Intel, just like PDVD DXVA and yours LAV DXVA.

How is that possible ?

[ksio89]

If you go to LAV Video properties inside MPC-HC and choose QS decoder (Intel QuickSync) and you tick UHD (4K), you will have HW acceleration for H.264 UHD (4K) files.

Man, that worked! There wasn't indeed any HW decoder selected in LAV video decoder within MPC-HC. I selected QS decoder and ticked UHD and HEVC boxed then played the UHD H.264 sample video again, and voilà, under status [H/W] Intel Quicksync showed up!

Playback was perfect: stable 29.97 fps, no dropped frames, low jitter and very low CPU usage! Then I selected DXVA2 native and CPU usage was even lower, dropped from around 30% to only 10% average.

I could also play the HEVC UHD sample @ 25 fps with HW acceleration activated, but as expected, only with DXVA2 native, but playback was stable as well. I'm very impressed with this results, Haswell HEVC decoder is better than I expected.

Man, I can only say thanks to you, MPC-HC and LAV developers! Seriously, I'm very grateful for your support here guys :thanks:

[P.J]

Look an older post of Ibius here in previous pages, around January with detailed directions about PowerDVD DSfilter use.

Strange, registered 'CLCvd.ax' but still no luck :confused:

[NikosD]

Did you change the merit from the default as Ibius said ?

With the default merit, the decoder is invisible to DXVA Checker.

[P.J]

Did you change the merit from the default as Ibius said ?

With the default merit, the decoder is invisible to DXVA Checker.

Ah, yas! :thanks:

[NikosD]

Yes, it works.

But this 8bit HEVC clip, still crashes Intel's DXVA (both native and cb)

http://demo-uhd3d.com/files/uhd4k/Hispasat_UHD_NAB-2014_HEVC.zip

This one crashes somewhere deep in the driver, those take some time to figure out. Presumably because its missing a reference frame and Intel expects it to be present.

Ideally, Intel would just fail to decode the frame instead of crashing completely, but from past experience, this is a common theme with Intel .. they crash when something is wrong.

Now that LAV has competition for hybrid HEVC DXVA decoding, I tried this clip with PDVD 15 DXVA.

It can't use HW acceleration like LAV/MPC, but on the other hand it doesn't crash the player like LAV/MPC.

It silently falls back to SW and decodes the clip fine.

[NikosD]

3) LAV/ MPC x64 versions are a little faster than the x86 versions, but not with a big difference.

LAV/MPC x64/x86 had some problems utilizing the GPU completely, sometimes they dropped the load a lot and iGPU frequency dropped too, resulting in lower performance e.g clips 2,4,6.

Optimizations needed for LAV DXVA in order to be competitive with PowerDVD 15 DXVA utilizing GPU at 100% all the time


because the DXVA decoder is in the driver it self it should come down to the buffering and not sure if this should be judge in benchmark mode.

It's confirmed.

Tried many difficult 2160p HEVC 8bit clips even from demo-uhd.com during normal playback and PDVD DXVA is far more smooth than LAV/MPC x64/x86.

LAV/MPC has many ups and downs in framerate during playback, dropping frames and stuttering.

On the other hand, play back using PDVD DXVA is smooth as butter, even in clips with higher bitrate than I would think.

The winner is clear in DXVA mode for Intel.

[P.J]

PDVD15 can't even play this file with DXVA2 (GTX960) but LAV: Hispasat_UHD_NAB-2014_HEVC

[NikosD]

As I wrote above, for Intel it's just the opposite.

It's a pesky HEVC file not being able to be decoded in DXVA by Intel due to Intel's drivers probably, but at least PDVD doesn't crash and falls back to SW.

What is the performance of PDVD DXVA on Nvidia 960 compared to LAV in general ?

[P.J]

Mostly equal, too bad PDVD doesn't support 10-bit with GPU.

[NikosD]

What do you mean ?

In CPU decoding playback of 10 bit HEVC clips is flawless.

For DXVA, OpenCL decoding I can't test it because Intel's Haswell GPU doesn't support it.

But if PDVD supports 10 bit SW decoding, how come it doesn't support it with DXVA ?

Weird. Are you sure ?

[huhn]

What do you mean ?

In CPU decoding playback of 10 bit HEVC clips is flawless.

For DXVA, OpenCL decoding I can't test it because Intel's Haswell GPU doesn't support it.

But if PDVD supports 10 bit SW decoding, how come it doesn't support it with DXVA ?

Weird. Are you sure ?

DXVA native with 10 bit doesn't really work with lavfilter too. copyback works fine.
EVR simple can't handle p010.

[P.J]

OpenCL doesn't work for Nvidia too.

[CPU: i3 4330, Astra HEVC 4K 10-bit, Decode performance]
PDVD15: 30.692 [20-57]
LAV x64: 39.375 [29-68]

About 4K HEVC, even PDVD14 supports 10-bit SW decoding but PDVD15 only supports 8-bit HW and 10-bit SW.

[NikosD]

OpenCL doesn't work for Nvidia too.


Have you installed latest Nvidia drivers with OpenCL 1.2 support ?

Try a 1080p file because OpenCL decoding doesn't work for 2160p on Intel too.

Maybe AMD cards can use OpenCL for 4K.


About 4K HEVC, even PDVD14 supports 10-bit SW decoding but PDVD15 only supports 8-bit HW and 10-bit SW.


As huhn correctly reminded me, there is no DXVA native decoder for 10 bit HEVC clips.

Only MPC Video Decoder can handle 10bit HEVC in DXVA native mode but with EVR-CP renderer not EVR and with compromises as Nevcairiel has said.

LAV uses DXVA-CB with EVR in order to accelerate 10bit HEVC.

PDVD DXVA doesn't support copy-back mode, so last chance is to try PDVD DXVA with EVR-CP renderer (I doubt it'll work)

[D2MAC]

Hi,

WIth this UHD 4K samples 10 bits :

https://www.transfernow.net/fr/330394i14jkx

DXVA2 COPY BACK work randomly.

[yingming]

Hi, a new version(2.0.3.2) of Strongene Lentoid HEVC decoder was released.

http://strongene.com/en/downloads/downloadCenter.jsp

[NikosD]

Hi,

WIth this UHD 4K samples 10 bits :

https://www.transfernow.net/fr/330394i14jkx

DXVA2 COPY BACK work randomly.

I can't see where the problem is.

I can play it with all decoders fine.

[Aleksoid1978]

Hi,

WIth this UHD 4K samples 10 bits :

https://www.transfernow.net/fr/330394i14jkx

DXVA2 COPY BACK work randomly.

Work perfect.

[NikosD]

Hi, a new version(2.0.3.2) of Strongene Lentoid HEVC decoder was released.

http://strongene.com/en/downloads/downloadCenter.jsp

Hello.

I see that you are from Beijing and you are developer, so my mind thinks that you are involved in a way with the development of Lentoid HEVC decoder.

I did some preliminary tests with the new decoder and saw that they fixed the two crashes - no crashes any more which is very good.

I'm about to run again the tests with the new decoder v.2.0.3.2 against the new LAV 0.65.

One question:

How can I change the number of threads from "Auto" to a specific figure ?

Whenever I do that from the properties of the decoder it reverts back to "Auto"

thanks

[NikosD]

[CPU: i3 4330, Astra HEVC 4K 10-bit, Decode performance]
PDVD15: 30.692 [20-57]
LAV x64: 39.375 [29-68]


This is not a very good comparison.

Can you test clips in DXVA native mode clips 2,4,6 where PDVD DXVA for Intel is faster than LAV DXVA ?

[nussman]

DXVA for 10bit clips on Intel Haswell?
This is not possible or did I miss something?

[yingming]

How can I change the number of threads from "Auto" to a specific figure ?

Whenever I do that from the properties of the decoder it reverts back to "Auto"


Do you mean that after you close the graphstudionext(we use this app to test performance), and open it again the Lentoid decoder filter threads was set to auto again?

Or only if you set threads to a specific number, it become auto immediately？

First situation is normal. But we never meet second situation before.

Thank you for your feedback.

[NikosD]

First situation is normal. But we never meet second situation before.



Yes, I meant the first situation, the second would be completely abnormal!

If the first situation is normal, then how could someone set the number of threads before using the decoder ?

Because even if you set it to 2 threads or 32 threads the actual performance is exactly the same.

The decoder doesn't seem to hold the value in the registry or a file.

And since you are here, I would like you to comment this:
http://forum.doom9.org/showthread.php?p=1718202#post1718202

It seems that in playback mode with any GPU (iGPU or dGPU) the playback in benchmark mode doesn't follow the decode in benchmark mode regarding performance.

Is there somewhere a bottleneck ?

Also I saw that v2.0.3.2 dropped a little the performance of x64 decoder for clips 4 and 5 (2160p), but not for the x86 decoder or the other clips.

Final request:
Add 10 bit HEVC support for the decoder x64/x86, in order to be perfect!

[filler56789]

Final request:
Add 10 bit HEVC support for the decoder x64/x86, in order to be perfect!

+1 :goodpost:

For the notes, I sent them that feature request on 2013-December-02. This is the reply they wrote me:

Dear sir,
the Lentoid filter will not support 10-bit HEVC decoding in the near future, thanks for your kind reminder.

The problem of course, is to know what they meant by "near future" :rolleyes:

[huhn]

sooner or later they have to.

broadcast is going to be 10 bit same for UHD BD.

even the windows 10 decoder can decode 10 bit HEVC and is pretty fast on top. sadly MF only...

[madshi]

Final request:
Add 10 bit HEVC support for the decoder x64/x86, in order to be perfect!
It's not about being perfect, it's about supporting UHD Blu-Ray, upcoming broadcasts, custom user encodes etc. I'm not sure if there will be a lot of 8bit HEVC sources. I think most HEVC sources will probably be 10bit. So I believe supporting 10bit is crucial.

[NikosD]

Exactly.

I said that in the context that it's a very fast and efficient decoder and lacks only 10 bit support in order to be perfect, meaning a complete ultra fast decoder in all cases.

Because if it was a slow decoder, nobody would really care if it has or not 10 bit support.

[huhn]

windows 10061 i3 4130 r9 270

decode mode

beauty

x86
microsoft MF
50 FPS 86% CPU
lav
22 FPS 92% CPU
lentoid
46 FPS 74%CPU
lentoid openCL
49 FPS 94% CPU 25% GPU

x64
microsoft MF
60 FPS 89% CPU
lav
49 FPS 72 % CPU
lentoid
48 FPS 78 % CPU

unexpected that MF wins so clearly.
but one thing is clear lentoid openCL is worthless on my setup.
about the important of x86 HEVC...
i think it's not important. a systme that can handle HEVC broadcast will be able to use x64. like newer computer games i guess it's time to drop x86 focus in the next years.

[nevcairiel]

FWIW, wrapping a MF decoder into a DS filter isn't very hard. The wmv9 mft decoder in LAV does exactly that - wrap the WMV9 MF decoder, and use it through LAV.
If that should be worth doing for the Microsoft HEVC decoder, I might do it. But not before Win10 is final.

[NikosD]

I haven't tested yet MS MFT HEVC decoder on newer CPU architectures - but only SSSE3 and SSE4.1

I'll test it with the final product.

The truth is that LAV x86 needs any help it can gets on all CPU architectures.

[nevcairiel]

Once content is actually available for HEVC, people might actually be interested in working on it. Its not like there is a company behind FFmpeg that pays developers to improve the HEVC decoder.
If it had just one paid full-time developer for the HEVC decoder, it would probably be running circles around all the competition, since all the basics are already in place, just a lot of optimizations missing - but alas, its open-source, and not a paid product.

In the long run, it'll probably end up the same as the H264 decoder. When H264 was new, a bunch of companies released decoders that performed well (at that time), to make a bit of quick money.
Over time, all these decoders died, since they were no longer maintained, and FFmpeg got faster and faster, and today its the fastest and most feature complete software decoder available to us.

[NikosD]

It seems to me that Lentoid x64/x86 decoders are free for the end user.

Also, competition is good for all, because if FFMpeg didn't have to catch the other decoders, probably the developers would get bored building something faster.

Anyway, LAV x86 is so slow, that is slower than anything else out there (PDVD x86, Microsoft x86, Lentoid x86), so the "easy" and free way to get a boost is the use of Microsoft's HEVC decoder built-in Win 10.

[nevcairiel]

People should really stop using 32-bit though, its always going to be slower, and be much harder to optimize for, and there is no good reason not to go 64-bit these days, not if you already use free players like MPC-HC which come in 64-bit anyway.

[NikosD]

Maybe, but you 're not the one who is going to tell people when and how are they going to upgrade their hardware and software.

Do you volunteer to upgrade their OS ?

Are you going to pay the new HW ?

And if you are talking about people, the truth is that people really don't care if the decoder or the SW product they use, is being built by paid or non-paid developers.

The main issue is if it's free for them, what is going to cost them and if it's fast - if it covers their needs.

[huhn]

the default linked version of mpc hc/be are both 64 bit.
so if you are not into this stuff you will get the 64 bit version this way.

the last thing that needs to change is codec packs people use these and they are usually not 64 bit . and there is no good reason for this anymore the last missing but still pretty special piece of software is reclock that it.

and people that can't use the 64 bit version have PCs that can't deal with HEVC in the first place.

the next thing is future hardware. where DXVA will dominate in the next couple of years making software decoder less important.

let's make this clear my current HTPC can't deal with a 100 Mbit UHD BD even if they increase the performance of software decoder by a factor of 2.

[ikarad]

1.ProRes-1080p@30fps-2Mbps



MPC x64 303/421/495 CPU 85%


What means theses numbers? 303? 421? 495?

[vivan]

min/avg/max fps

[ikarad]

min/avg/max fps

Thanks.

[P.J]

I have tried Win10 10061 with GTX960 and can't play 10-bit videos but 8-bit videos work fine =T

[huhn]

just a small update about UHD code.

beauty
UHD code 11.4
31 FPS 85 % CPU

MS MFD 10122
59 FPS 86% CPU

smooth playback is still out of question with UHD code the mircosoft decoder is easily 90 % faster.

[D2MAC]

just a small update about UHD code.

beauty
UHD code 11.4
31 FPS 85 % CPU

MS MFD 10122
59 FPS 86% CPU

smooth playback is still out of question with UHD code the mircosoft decoder is easily 90 % faster.

Hi,

MS MFD 10122 ? (samples ? 8 bits / 10 bits ? mkv container / .ts container ?)

I've a GTX960 and WINDOWS 10 - 10122 preview.

how can i use MS MFD ?

Windows Media player, can't open my HEVC UHD 4K samples (.ts container)

So windows 10 can't open HEVC 4K UHD in .ts container ?

WIth MPC-BE player it's ok or MPC-HC with lav filters. (with DXVA COPY-BACK - 2% CPU)

[huhn]

Hi,

MS MFD 10122 ? (samples ? 8 bits / 10 bits ? mkv container / .ts container ?)

I've a GTX960 and WINDOWS 10 - 10122 preview.

how can i use MS MFD ?

Windows Media player, can't open my HEVC UHD 4K samples (.ts container)

So windows 10 can't open HEVC 4K UHD in .ts container ?

WIth MPC-BE player it's ok or MPC-HC with lav filters. (with DXVA COPY-BACK - 2% CPU)
mircrosoft media foundation can be test with dxva checker. and it is used with the "video" application.
mp4 and mkv with hevc work with this decoder/splitter.
and this is about software decoding not hardware.

[D2MAC]

mircrosoft media foundation can be test with dxva checker. and it is used with the "video" application.
mp4 and mkv with hevc work with this decoder/splitter.
and this is about software decoding not hardware.

.mp4 and .mkv

but .ts container ?

[huhn]

.mp4 and .mkv

but .ts container ?

i don't know sorry.

[D2MAC]

i don't know sorry.

Hi,

.ts in HEVC 10 bits to download here :

http://demo-uhd3d.com/fiche.php?cat=uhd&id=83

download it and tell me if it's work with Microsoft codec ?

[P.J]

TS doesn't work in Win10

[D2MAC]

Hi,

Does Microsoft plan to support .ts HEVC in WIndows 10 final ?

They claim that they support HEVC in WINDOWS 10.

But no, not full ....

Without .ts support is disapointing

.ts is a popular container in Broadcast.

Microsoft is aware about .ts / HEVC problem?

Same problem in VLC with HEVC + .ts

[huhn]

and they support HEVC up to main10.
there is simply no TS demuxxer with HEVC support. but this doesn't change that windows 10 fully supports HEVC.

[D2MAC]

and they support HEVC up to main10.
there is simply no TS demuxxer with HEVC support. but this doesn't change that windows 10 fully supports HEVC.

It's annoying.

there is a lot of .ts UHD 4K samples.
lav filters)
And broadcast use .ts.

they should add a .ts demuxer for HEVC like MPC-BE/MPC-HC

there a lot .ts HEVC samples :

example (found on the web) :

HEVC 10 BITS .ts container download link :

http://www.demo-uhd3d.com/files/uhd4k/Channel_4K.ts

http://www.demo-uhd3d.com/files/uhd4k/Roland_Garros_MonFils_2014.ts

http://www.demo-uhd3d.com/files/uhd4k/Samsung_Fashion_Show_Milan_UHD.ts


HEVC .mp4 container :

http://www.demo-uhd3d.com/files/uhd4k/LG_nuit_villes.mp4

HEVC .mkv container :

http://www.demo-uhd3d.com/files/uhd4k/IMAX_PreShow.mkv

[P.J]

Who's gonna use windows media player again?

[NikosD]

It's been 8 months since my first comments about Optimus technology and HEVC decoding here:
http://forum.doom9.org/showthread.php?p=1694992#post1694992

Unfortunately, not a lot have changed since then.

I tried on a clean Win 8.1 Pro x64 - Core i5 4200M - dGPU Nvidia 740M - iGPU HD 4600 with drivers 4061 for Haswell and 347.88 for Kepler with no better luck.

I did my benchmarks again with latest LAV x64 0.65.9 in DXVA native mode and DXVA Checker 3.4.1 in Decode Benchmark mode for all 8 HEVC clips 1080p & 2160p.

The tests were done in battery mode and balanced power profile.


1.ProRes-1080p@30fps-2Mbps


LAV CPU 113/165/206 CPU 89%

LAV DXVA 740M 105/147/192 CPU 26%

LAV DXVA HD4600 74/110/131 CPU 45%




2.Elephants Dream-1080p@24fps-1.7Mbps


LAV CPU 54/168/289 CPU 76%

LAV DXVA 740M 0/145/260 CPU 28%

LAV DXVA HD4600 50/113/160 CPU 42%




3.Big Buck Bunny-1080p@60fps-2Mbps


LAV CPU 88/204/288 CPU 89%

LAV DXVA 740M 0/157/276 CPU 23%

LAV DXVA HD4600 83/133/160 CPU 44%




4. Girls -1080p60 fps -11Mbps


LAV CPU 75/123/175 CPU 87%

LAV DXVA 740M 73/115/173 CPU 27%

LAV DXVA HD4600 59/95/126 CPU 48%




5.Beauty-2160p@30fps-12.3Mbps


LAV CPU 28/35/39 CPU 73%

LAV DXVA HD4600 22/27/30 CPU 33%

LAV DXVA 740M SW fall-back



6.Fitness-2160p@30fps-8Mbps


LAV CPU 35/45/70 CPU 90%

LAV DXVA HD4600 27/33/43 CPU 61%

LAV DXVA 740M SW fall-back



7.Ducks-2160p@50fps-4Mbps


LAV CPU 45/51/52 CPU 95%

LAV DXVA HD4600 31/36/38 CPU 35%

LAV DXVA 740M SW fall-back



8.Gravity - 2048x858@24fps-5Mbps-10bit

The hybrid decoders of Intel and Nvidia don't support 10 bit HEVC decoding, so all decoders work but in SW fall-back mode.

Intel is still crashing with P010/P016 output enabled, while Nvidia has fixed the bug.



Comments:

1) Nvidia decided to cut the HEVC decoder above 1080p, at least on Kepler, so everything above that resolution falls back to SW.

2) As you can see from the minimum figures, Nvidia still has problems decoding HEVC streams with its hybrid decoder.
The issue remains unresolved for almost a year now, so I wouldn't trust that decoder for HEVC decoding - I don't know if Maxwell drivers are any better than this.

3) Intel is still using 100% GPU Load but with lower than max clock. Only 800-850MHz of 1150MHz max.
Even in High Performance mode or with the laptop plugged-in, the clock doesn't go any higher.

4) Nvidia, on the other hand has a full clock of 1058MHz but with low GPU utilization in Balanced mode ~60-65%
In High Performance or in plugged-in mode the utilization goes up to 90-100% and the performance rises.

This is definitely an improvement compared with older drivers, but the incompatibility problems and the cut-off resolution of 1080p, are far more serious issues.

[NikosD]

Well, this time it looks like they did it right, we only have to wait for Nevcairiel to build the 10bit decoder and see the performance on that.

Based on the previous first generation 4K H.264 decoder (VP5), they didn't manage to even reach 4K30fps.
VP5 is a 4K26-28fps decoder.

So, we didn't know exactly what to expect from Nvidia.

Kepler (600, 700 series)'s VP5 can easily reach 4K35fps~4K40fps.
Only GT520/610(GF119) can't reach 4K30fps, even GT520M(GF119 as well) can reach 4K31fps, my friend tested before.

I did a test to check those claims about Kepler's VP5 performance regarding 4K H.264 compared to Fermi's VP5, especially GT610.

Well, there is a small difference of about 10%, not worth to mention I think.

In playback mode with scaling 1280x720, Fermi was a little faster, probably an incompatibility of Kepler with the specific file.

I used same drivers 347.88 for Kepler and Fermi, LAV x64 DXVA 0.65.9 in DXVA native and DXVA Checker 3.4.1 in both Decode Benchmark mode and Playback benchmark mode.

The laptop with Optimus technology was used in both battery - balanced profile mode and plugged-in / high profile mode with no difference in results.

Decode mode


1. Chimei-inn-3840x2160@30fpsRef3-50mbps

Kepler 740M VP5 31/34/36

Fermi GT610 VP5 28/31/32


2. RED-4096x2160@24fps-36Mbps


Kepler 740M VP5 30/36/41

Fermi GT610 VP5 28/32/36



Playback mode


1. Chimei-inn-3840x2160@30fpsRef3-50mbps
1280 x720


Fermi GT610 VP5 25/28/30

Kepler 740M VP5 22/26/29



2. RED-4096x2160@24fps-36Mbps
1024 x 768

Kepler 740M VP5 25/31/40

Fermi GT610 VP5 25/29/32

[NikosD]

Socketed Broadwell for desktop is out:
http://www.tomshardware.com/reviews/intel-core-i7-5775c-i5-5675c-broadwell,4169.html

Unlocked CPU with lower clocks than top Haswell, only 65W TDP in 14nm and amazing GPU performance over HD4600 (more than double) by Iris Pro 6200@1150MHz vs HD4600@1250MHz.

In games it has the speed of R7 250 with a 12Watt GPU (!)

Hybrid HEVC(8bit/10bit) and VP9 decoding using CPU/GPU and overall a lot faster transcoding than Haswell (in theory)

CPU performance of Core i7 5775c ~ Core i7 4790.

Two problems:

1) Skylake is too close for release, that's why Intel decided to cannibalize Broadwell for desktop with only 2 models (!)
Core i7 5775C and Core i5 5675C

But Skylake for desktop could be a disappointment because it will have half-speed of Braodwell in GPU (24EUs vs 48EUs) and I read that no AVX-512 version will be included in desktop Skylake (only Xeons)

2) No LGA1150 chipset 8-series support for Broadwell. Only 9-series.

ARE YOU SERIOUS INTEL ?

They should support all LGA1150 motherboards, because only 2 SKUs are out and one of them is a Core i5.

Nobody with a 4790K will "update" to Broadwell, with lower CPU performance besides serious overclockers and nobody with a 9-series chipset will put a Core i5 Broadwell on it.

[nevcairiel]

Broadwell is not very interesting. Hybrid decoding is a stop-gap measure without much merit. Skylake will be out in a couple month (August seems to be the target) and offer full hardware decoding, so its really wiser to just wait.
I certainly will update my HTPC with a Skylake CPU later this year, although probably not the very high-end, its a HTPC afterall.

[NikosD]

August will be presented, but about October will be sold (!) as I read somewhere.

Skylake is not very interesting too.

About same CPU/GPU performance with Haswell with no AVX512.

Only fixed - function HW will be interesting but I will not buy a new board and new RAM DDR4 only for that.

We need AMD back to competitive CPUs and we need AMD NOW !

[P.J]

Only up to 4Kp30...

[NikosD]

No, it's not accurate just fail safe.

My HD4600 can already decode much better than that, with low bit rates.

I guess it can reach 4K60fps but with much lower than BluRay UHD bitrates.

[NikosD]

AMD's Carrizo is announced with HEVC HW decoding, but no performance figures yet.
It's only for laptops, though.

The milestone is the HSA 1.0 compliance, something that could potentially change the whole CPU/GPU subject.

The new HandBrake will have VCE support and Carrizo is faster in transcoding 1080p H.264 than Broadwell Core i5 - 5200U, according to AMD.

http://arstechnica.com/information-technology/2015/06/sixth-time-lucky-amd-details-the-carrizo-apu/1/

[foxyshadis]

Nothing supports AVX-512 as it is, so exclusion of it isn't exactly very damning, while the 24 EU count isn't confirmed anywhere and can't be considered anything other than wild ass guessing, at this point. Best anyone can do is sit tight and wait a month for real announcements before getting their hopes up (or dashed).

AMD just released HEVC hardware decoding. I'm looking forward to the benches from that.

[huhn]

even if they stay at the same EU number they can be still a lot faster.

[NikosD]

You are both wrong, but I won't report my internal source from Intel to prove you wrong.

Skylake for desktop will be released with 24EUs.

About performance, it's clear that 48 EUs with 128MB eDRAM is at least 2x faster than 24EUs.

Comparing to Haswell will be at most 20% faster, which is not worth to mention it.

[foxyshadis]

I don't doubt they'll be launched with 24 EUs -- launching from the bottom seems to be Intel's strategy ever since Merom -- but I don't expect it to stay that way for long. Broadwell started at 24 and now has 48, after all. It really seems like Intel is deliberately drawing out its performance part releases to let yields and cost catch up, instead of releasing $2000 chips and hoping for a market.

[blaubart]

The good news is that as a separate side-project I've been working on reducing the CPU usage of Copy-Back, and assuming a direct decoder -> output chain (no software deint, no format conversion, clean nv12 decode and nv12 output, or p010 decode and p010 output), I could reduce the CPU usage on my NVIDIA system by up to 50%, and additionally the benchmarked performance is now 99% the same as DXVA2-Native benchmarks. These gains may be a bit smaller on Intel or AMD GPUs, but they will still be quite noticeable.

I may also add a P010 decode -> NV12 output path using these optimizations, as that may be a common use-case for now when viewing 10-bit content on many renderers.

May I say thanks to nevcairiel,
though my 4K testTV LG 42UB820V probably not 10-bit at all, anyway, I got per DVBViewer (live TV) the Astra UHD Channel (50fps 3840x2160 HEVC) to run smooth and fluently (CPU 7%, GPU 30%) (Intel i7-4790K, Nvidia GTX 960).

Your "GTX960 @ dxva2cb direct" improvements made this possible for the very 1st time (even with an active DVBSource). Thank you again!
If you speak German... (http://www.dvbviewer.tv/forum/topic/55897-quo-vadis-htpc-richtung-hevc/?p=428725)

[P.J]

http://www.techpowerup.com/213575/amd-fiji-silicon-lacks-hdmi-2-0-support.html

[huhn]

the nano had a chance to be a really good HTPC card but something is totally wrong with AMD how can they miss that a lot of people buy nvidia 9xx cards just for the HDMI 2.0 cnnector.

[iSunrise]

the nano had a chance to be a really good HTPC card but something is totally wrong with AMD how can they miss that a lot of people buy nvidia 9xx cards just for the HDMI 2.0 cnnector.
They buy it since it's perfect for a HTPC, has good performance/efficiency and it has a fixed-function decoder for both H.264 and H.265/HEVC. Last but not least, yes, it also has HDMI 2.0/HDCP 2.2. It's not just because of the HDMI 2.0 connector, but that is part of the equation to support all upcoming 4K content.

Fiji was primarily build for PC gaming (DP only 1:16) and AMD just didn't have their decoder/display block updated for some reasons that are currently unknown.

[huhn]

most people don't buy a 960 for an HTPC. madVR user do but let's get real this is still a gaming card not a amd 5450/6450 media card.

you can't use TVs as your UHD display now with the nano and that's a very huge issue. normal UHD monitors are pretty crappy and small UHD TV are pretty cheap and have a way better PQ and AMD totally missed that market.

[iSunrise]

most people don't buy a 960 for an HTPC. madVR user do but let's get real this is still a gaming card not a amd 5450/6450 media card.
Well, it was you who claimed that "a lot of people buy nvidia 9xx cards just for the HDMI 2.0 cnnector." Since HDMI doesn't make much sense for PC, but only for TV I was just pointing out that HDMI 2.0 does only make sense for HTPC use. For everything else DP is the perfect solution, since even the highest-end monitors all do DP perfectly fine (Eizo CG series, NEC, etc.).

you can't use TVs as your UHD display now with the nano and that's a very huge issue. normal UHD monitors are pretty crappy and small UHD TV are pretty cheap and have a way better PQ and AMD totally missed that market.
While that is true, it not totally unusable. With HDMI 1.4a you can playback movies with 4K and up tp 30 fps perfectly fine, you just cannot use specific UHD-content like Netflix or other content providers that require HDMI 2.0 and HDCP 2.2 as copy protection. But you can playback everything directly from your PC with 4K and <=30fps just fine. So the nano still has a place for the HTPC market, but it's limited.

4K is still in it's infancy, there isn't even a lot of content for it right now. The nano is limited here, but it's not like it's totally unusable. Yes, it would be perfect if it had HDMI 2.0/HDCP 2.2.

It's nothing new, for several years now, AMD is behind when implementing state-of-the-art display/encoder/decoder blocks and that is a very sad thing. In the past, AMD was always at the forefront, way outclassing NV in video, but that time has been long gone.

[nevcairiel]

I think his point is that many people actually use small TVs (around 32-40") as PC displays, and they would greatly appreciate HDMI 2.0 for 4k@60.

[iSunrise]

I think his point is that many people actually use small TVs (around 32-40") as PC displays, and they would greatly appreciate HDMI 2.0 for 4k@60.
Most of the displays I know don't even support HDMI 2.0, yet, including the Philips 40" 4K-TV (VA panel), which is often used and is very popular (for TV and PC use).

But yes, I certainly agree, the nano will not have a very long life in it, I think we can safely say that. We can only hope that AMD knows that this is a serious limitation for a HTPC card and will not take too long to come up with a better successor.

[huhn]

even the last years samsung support 4:4:4 at 60 hz.

and yes some TV still have old 10 gbps chip for example all sonys and most Phillips if not all Phillips.

here an example: http://www.hdtvtest.co.uk/news/ue48h6400-201411133951.htm

EDIT: Phillips was well know for a good PC modes but this totally changed with the UHD TVs same for sony.

let's see when the totally pointless TP vision bashing starts.

[Yups]

You are both wrong, but I won't report my internal source from Intel to prove you wrong.

Skylake for desktop will be released with 24EUs.

About performance, it's clear that 48 EUs with 128MB eDRAM is at least 2x faster than 24EUs.

Comparing to Haswell will be at most 20% faster, which is not worth to mention it.


BDW GT3e won't be 2x faster unless GT2 Gen9 can't outperform a GT2 Gen7.5. Also AVX512 is uninteresting for us. Look to AVX2. Barely a 5% improvement in x264 and beside this basically no other productive consumer apps are available who can really make use of AVX2. No AVX512 is no loss and given that AVX2 was a bigger issue regarding tenmperature check tools like Prime95 it might be better.

[NikosD]

BDW GT3e won't be 2x faster unless GT2 Gen9 can't outperform a GT2 Gen7.5.


I can't follow your math here.

GT2 Gen7.5 has 20 EUs and GT2 Gen9 has 24 EUs, so GT2 Skylake will be a little faster than GT2 Haswell.

But Broadwell desktop has 48 EUs with 128MB super fast eDRAM, so it will be 2x faster than Skylake GT2, assuming that internal architecture of Skylake will not cover the eDRAM additional performance.


Also AVX512 is uninteresting for us. Look to AVX2. Barely a 5% improvement in x264 and beside this basically no other productive consumer apps are available who can really make use of AVX2. No AVX512 is no loss and given that AVX2 was a bigger issue regarding tenmperature check tools like Prime95 it might be better.

x264 is yesterday, it has practically stopped developing any more.

x265 is today and tomorrow and has already improvements of around 20% using AVX2 - IIRC - and it will be further optimised in the near future.

I'm not sure though if AVX512 would help even more.

Anyway, AVX512 just like AVX2 has more uses than HTPC and I always consider a more general view regarding my CPU upgrades.

[Yups]

I can't follow your math here.


Just look into some of the BDW GT3e reviews. There is roughly a 2x gap to HSW GT2. Your thinking is flawed. 2x of the EUs doesn't result in 2x performance because of other limitation, otherwise we would have seen a 2x gap on HSW GT2>HSW GT3(e).....it wasn't the case.



GT2 Gen7.5 has 20 EUs and GT2 Gen9 has 24 EUs, so GT2 Skylake will be a little faster than GT2 Haswell.


You have to assume that beside the rather small EU increase nothing changed regarding efficiency or other aspects of the GPU. For example there is a 1.5x increase in texture sampler from Gen 7.5 to Gen8 (GT2). You also have to assume that Intel didn't improve anything performance relating from Gen8 to Gen9. I highly doubt HSW Gen7.5 GT2 can match a Skylake Gen9 GT2.

[NikosD]

Just look into some of the BDW GT3e reviews. There is roughly a 2x gap to HSW GT2. Your thinking is flawed. 2x of the EUs doesn't result in 2x performance because of other limitation, otherwise we would have seen a 2x gap on HSW GT2>HSW GT3(e).....it wasn't the case.



I think your thinking is flawed.

Almost all of the comparisons I have seen they compare a GPU of 1100MHz (5675c) or 1150MHz (5775c) with a Haswell GPU of 1200MHz or most of the times 1250MHz.

So, yes if you compare a 48EUs 1100 MHz vs a 20 EUs 1250 MHz GPU the
difference couldn't be exactly doubled or even more than doubled, but we are playing with figures here around very small percentages near to x2+ performance.


You have to assume that beside the rather small EU increase nothing changed regarding efficiency or other aspects of the GPU. For example there is a 1.5x increase in texture sampler from Gen 7.5 to Gen8 (GT2). You also have to assume that Intel didn't improve anything performance relating from Gen8 to Gen9. I highly doubt HSW Gen7.5 GT2 can match a Skylake Gen9 GT2.

Since sandybridge and the first iGPU of Intel, the most critical parameter regarding GPU performance was always the number of EUs rather than internal architecture.

So I expect a GT2 Skylake to be around 20% faster than GT2 Haswell.

[huhn]

the 750 ti is using 6 watt so i can easily assume they measured the GPU only.

i have an r9 270 which can't dynamically change the clock speed of it's Vram only max or min which is 150/1400(nearly all other r9 270 should be able to do that). and i get a whole system of about 40 watt at the power supply! this doesn't match at all with this chart.

i will grab my power consumption meter and will check again. but there results looks fishy and they could be easily a problem with the new driver.

[wanezhiling]

http://tpucdn.com/reviews/AMD/R9_Fury_X/images/power_bluray.gif

AMD still haven't fixed their ridiculous power consumption when decoding Blu-ray, 66W is just wasteful.
:p
66 is not a high number, indeed.

[huhn]

:p
66 is not a high number, indeed.

these are really high numbers for GPU only but this german test shows fury x beats the 980 ti in term of power efficient hardware decoding but lose to the 980 for FHD and wins at UHD.

the hardware decoder of this card looks like a boss.

http://www.computerbase.de/2015-06/amd-radeon-r9-fury-x-test/8/

so i take all these numbers just as "numbers" right now.

over 100 watt for DXVA decoding there base test system must be very wasteful which is ok to make sure the bottleneck is the GPU not the CPU in a test.

[P.J]

Mine uses only 8W :D

I believe Intel cpus use even less than Radeon graphic cards :p

[wanezhiling]

http://i4.tietuku.com/05398b87a064af75.jpg

r9 380 (r9 285), tonga already supports hevc?????:eek:

[P.J]

http://i4.tietuku.com/05398b87a064af75.jpg

r9 380 (r9 285), tonga already supports hevc?????:eek:

No 10-bit and It should be hybrid.

[P.J]

http://www.game-debate.com/news/?news=17350&graphics=GeForce%20GTX%20950%20Ti%202GB&title=Nvidia%20Looks%20To%20Be%20Readying%20A%20GeForce%20GTX%20950%20Ti%20Graphics%20Card

[NikosD]

Added a post to article's comments that they must use latest nightly build of LAV filters in order to use 4K H.264 DXVA decoder.

The problem with all these reviews is that they don't say a lot about HTPC use of such cards.

Anandtech.com used to dedicate a separate article for HTPC use.

They could probably do it for Nano card.

[NikosD]

I was already disappointed by Skylake.
That was the final hit.

So, it's only Nvidia then.
BAD thing.

Let's hope that by the time BluRay UHD and other HEVC 10bit content become popular, all three will have decent 10 bit HW decoders.

[wanezhiling]

first 4k h264 card -> ancient gt520
first 8bit h265 card -> gtx960
first 10bit h265 card -> gtx960

[P.J]

Intel http://forums.mydigitallife.info/vb4_style/smilies/poop.png

[wanezhiling]

http://img1.mydrivers.com/img/20150710/dcbf81f3611847fea6403a3838977b0f.png

i7-6700K

[NikosD]

Impressive, but is 10bit HEVC supported by hybrid or pure HW acceleration ?

To be 100% sure, we have to wait for a review.

[P.J]

It should be the same as Broadwell.

[huhn]

looks like MS MF HEVC decoding with windows 10240 is broken. did someone have the same problem?

[CruNcher]

With H.264 it wasn't as clear but this time with H.265/HEVC the decoder in lavfilter is blowing away every commercial solution so far in a combined 10 Bit Software 8 Bit Hardware workflow Performance test.

Especialy vs the x64 version Commercial solutiions currently see no real land :)

Elecards x86 Demo has no chance even the performance of their x86 code path is not up to the 32 bit part of the lavfilter used decoder crazy :)

http://www.elecard.com/en/products/end-user-software/playback/hevc-decoder-plugin.html

[NikosD]

Hi there.
It's been a long time.

Where have you been ?

To be honest, the situation is exactly the opposite.

It's the first time that LAV/FFmpeg HEVC decoder is awfully slow at x86 version and the HW DXVA performance of PDVD 15 is a lot better as a total expereience.

x86 H.264 SW decoder didn't have such problem.

PDVD HW DXVA is faster, smoother and with better error handling at least with Intel GPUs and their hybrid approach.

[CruNcher]

on WDDM 1.1 it looks way different here for Nvidia :)

and the Software x86 paths of Elecard and co are even slower and jitter more overall

Benchmark is something different then a Playback test where you CPU overhead is crucial for output and stable realtime render frametimes

also im testing on the lower end i5-2400 where every cycle counts especially for the 4k 10 bit Playback test :)

im currently using mainly 1 test scene done by Atemes H.265 Broadcast Software Encoder one of the Samsumg Demo Streams aired on Astras Hevc Demo Channel

Samsung_Fashion_Show_Milan_UHD.ts

http://demo-uhd3d.com/fiche.php?cat=uhd&id=44

i find the beauty sample a little strange

http://forum.doom9.org/showpost.php?p=1730251&postcount=19407

Here is a Real Life x64 playback test of the 10 bit stream it's also a compare of the Efficiency of the Render core of both players @ the same time

http://forum.doom9.org/showpost.php?p=1730299&postcount=2638


the results of PDVD15 are really not so good either with HAM or DXVA for this specific config, but Nvidia seems to still optimize on the Driver ;)

Powerdirector 13 Demo has the 64 bit decoder will try that next

Arcsoft seems to have left the Consumer Video PC Buisness to concentrate on Mobile and Pro Solutions it seems :(

So it looks like this for the current updated Cyberlink Decoder Release

32 bit = 7 june 2015
64 bit = 5 January 2015

The 32 Bit DXVA result on the GTX 970 awefull playback experience :(

but still better then the Software Part of Cyberlinks decoder alone :)

http://abload.de/thumb/brakeups09j2u.png (http://abload.de/image.php?img=brakeups09j2u.png)

32 Bit Software Decoding 64 bit is virtually the same result

http://abload.de/thumb/softwaredecodetrbsl3.png (http://abload.de/image.php?img=softwaredecodetrbsl3.png)

Maybe there are some problems with MPC-HC and its handling of colorspace though between Cyberlinks Decoder output and the Nvidia Driver as Lav outputs in NV12 by default

Lav Filter x86 Software Decoding basicly the same MT under utilization result of Cyberlinks x86 as well as x64 code path

http://abload.de/thumb/lavvideo32tnbq8i.png (http://abload.de/image.php?img=lavvideo32tnbq8i.png)


And now Lav Filter x64 10 bit Software Decoding (Amazing) no MT brakeups everything absolute jitter free :)

http://abload.de/thumb/supercleani6kmy.png (http://abload.de/image.php?img=supercleani6kmy.png)

@NikosD

What is the Decoding result with Intels Hybrid mode of that 4K@25Fps 10 Bit 50 mbps clip when you further restrict it to 4 cores ?


Intel's hybrid decoder doesn't support 10 bit HEVC decoding, so all decoders work but in SW fall-back mode with the exception of Lentoid OCL which crashed.

ahh this is still the case ?

[NikosD]

From your post you seem to forget two very interesting decoders.

Lentoid x86/x64 and Microsoft's x86/x64 HEVC decoder.

Lentoid x86 is the fastest x86 HEVC decoder, with a serious limitation of no 10bit decoding.

MSFT depending on the CPU architecture is also very fast and for x86 generally faster than LAV x86.

For DXVA decoding, PDVD 15 is generally smoother than LAV in real life playback.

When I use the word faster, I mean benchmarks which is translated to lower CPU usage in playback normal mode.

When I use the word smoother I always mean real time playback.

Regarding OCL (HAM) PDVD 15 and other OpenCL HEVC decoders are not as good as DXVA decoders.

So, as I wrote above I use PDVD DXVA mode for hybrid real time playback experience.

About your question, generally speaking using a hybrid DXVA mode for decoding isn't affected by the number of cores when I use a CPU like core i7-4790.

Hyperthreading has a real impact in CPU decoding, not hybrid.

Your question is not valid because you are asking about hybrid decoding of a 10 bit clip, which means SW decoding, because there is no support of 10bit decoding in HW or shaders for Haswell

[CruNcher]

im only considering 10 bit supporting decoders that can decode that test clip if they can't they useless ;)

so lentoid and uhdecode are not part of any testing because 8 bit Hardware Decoding (doesn't really matter if Copyback,Native or CUVID only some exceptions currently as with the beauty clip) is absolutely efficient with Lav Filter alone

PowerDVD 15 main Application is currently broken for me (Maxwell V2) i can't test that get no output currently most probably a Nvidia Driver issue with their renderer

The good old Arcsoft Theatre 6 still works without any issues

MSFT Decoder is no option because WDDM 1.1 testing only most probably it's based on Mainconcepts HEVC SDK anyway ;)

yeah i saw your post about intels hybrid decoder for haswell has no 10 bit support so you cant test that.

[huhn]

the microsoft decoder can decode 10 bit.

[CruNcher]

the jitter issue seems only really fixable hardware accelerated in different configuration with Nvidia Cuvid on that beauty sample

http://abload.de/thumb/fixed0su4v.jpg (http://abload.de/image.php?img=fixed0su4v.jpg)

[P.J]

How did you enable that information?

[CruNcher]

Ctrl+j

Crazy Stuff :)

DXVA

http://abload.de/thumb/3xperformancelossgns73.png (http://abload.de/image.php?img=3xperformancelossgns73.png)

CUVID/Copyback

http://abload.de/thumb/3xperformancewinuwsw2.png (http://abload.de/image.php?img=3xperformancewinuwsw2.png)

x64 Software

http://abload.de/thumb/softwaregyuob.png (http://abload.de/image.php?img=softwaregyuob.png)

[Yups]

So, yes if you compare a 48EUs 1100 MHz vs a 20 EUs 1250 MHz GPU the
difference couldn't be exactly doubled or even more than doubled, but we are playing with figures here around very small percentages near to x2+ performance.


Couldn't be? Maybe re-check the reviews, it is pretty much a doubling. Some games a bit more, some a bit less but overall a doubling.



Since sandybridge and the first iGPU of Intel, the most critical parameter regarding GPU performance was always the number of EUs rather than internal architecture.


There was a pretty big overhaul with Gen7 (especially GPU Computing), basically twice the computing power. At the same clock there was a 50% improvement in games with 33% more EUs over Sandy Bridge, with current drivers probably much more. Not to mention that EU increases don't scale linear because of other limitations like bandwidth. A 20% EU increase alone is unlikely to give 20% more extra performance. Gen8 did with the help of other enhancements in the architecture (like more texture samplers). As a result SKL GT2 with 24 EUs is most likely way more than a doubling over Sandy Bridge with 12 EUs.


So I expect a GT2 Skylake to be around 20% faster than GT2 Haswell.

I expect a 30% improvement which lowers the difference to ~75%. And even with your 20% expectation a doubling isn't really accurate.

[NikosD]

looks like MS MF HEVC decoding with windows 10240 is broken. did someone have the same problem?

I was stuck at build 10049 with my second backup system because Windows update was broken in both slow and fast ring.

So, today I decided to do a clean install of latest 240 build (RTM ?).

It seemed to me like a very stable and bug free release especially when compared to 049 after a few hours of work.

Edge is very fast, but is still missing features compared to IE.

Anyway, regarding MS MFT HEVC decoder, it didn't work.

I used latest DXVA checker and nightly LAV filters and although DXVA checker enumerated the MS HEVC decoder, it didn't show up with various .mkv, .mp4

I'm guessing incompatibility with DXVA checker (didn't try WMP) or LAV splitter or a Microsoft's bug.

[sneaker_ger]

Isn't the Microsoft decoder Media Foundation only? Then it cannot connect to DirectShow splitters like LAV.

[NikosD]

You are right, MS HEVC decoder is MFT only.

[huhn]

so my system isn't broken good to know.

[NikosD]

Latest Nvidia drivers 353.30 are still WDDM 1.3 - DX 11.x for Fermi.

Still no WDDM 2.0 - DX12 drivers for Fermi, even for Win 10 RTM

[huhn]

they may never come they are relatively old cards. the current windows 10 driver is 353.50 maybe kepler+ only.

[aufkrawall]

Nvidia officially have announced DX12 support for Fermi GPUs.

[huhn]

http://blogs.nvidia.com/blog/2015/05/15/dx-12-game-ready-drivers/

ok no questions.

feature level is already a problem for kepler cards so real dx 12 needs the new cards

[P.J]

Latest Nvidia drivers 353.30 are still WDDM 1.3 - DX 11.x for Fermi.

Still no WDDM 2.0 - DX12 drivers for Fermi, even for Win 10 RTM

Then the Windows 10 driver is still WDDM 1.3 for GTX900 series?

[NikosD]

http://blogs.nvidia.com/blog/2015/05/15/dx-12-game-ready-drivers/

ok no questions.

feature level is already a problem for kepler cards so real dx 12 needs the new cards
You have misunderstood what Microsoft did and wanted for DX12.

It's the first time that Microsoft allowed for old and very old HW like Fermi to support the basic very crucial features of the new DX12 API like the low CPU overhead.

That doesn't mean of course that Fermi could support every feature of DX12, just the basic, very essential core features.

That's why MS implemented the feature levels in order to add features on the basic core, for advanced HW.

So, when we are talking about Fermi's or Kepler's DX12 support, we mean the basic "core" elements of the new API but with feature level of DX11.x

Nvidia has promised DX12 support for Fermi and I' m waiting for it after the official release on 29th of July.

[huhn]

directx 11 has feature levels too and most nvidia cards doesn't support them up to 11.2. AMD does support 11.2. there is even feature level 11.3.

older directx version where pure software updates so this is nothing new with dx 12.
feature level 11.2 is simply not used so it doesn't really matter but it's a bad after taste.

maxwell is dx12.1 not 12.0 like kepler.

dx 12 software mode and dx 12 full are to different things and it will play a rule in the feature. maybe for video decoding/presentation maybe not.

if windows 10 gets window mode with 10 bit i wouldn't be shocked a direct x 12.1 card is needed.

[NikosD]

directx 11 has feature levels too and most nvidia cards doesn't support them up to 11.2. AMD does support 11.2. there is even feature level 11.3.

older directx version where pure software updates so this is nothing new with dx 12.
feature level 11.2 is simply not used so it doesn't really matter but it's a bad after taste.

maxwell is dx12.1 not 12.0 like kepler.

dx 12 software mode and dx 12 full are to different things and it will play a rule in the feature. maybe for video decoding/presentation maybe not.

if windows 10 gets window mode with 10 bit i wouldn't be shocked a direct x 12.1 card is needed.

Your post is completely wrong and misleading.

You seem to confuse basic things about new DX12 and although I wrote that before, you obviously didn't bother reading more about this new API before you write another post.

There is no such thing as "dx12 software mode"

The basic, core features of DX12 like low CPU overhead even in single threaded, multi - thread and multi-GPU are features of the new DX12 API that Microsoft wanted to be used by as much older HW as possible and of course they are not "software mode" features, they are part of the new WDDM 2.0 driver.

They use less CPU instead of increasing it, like if they were executed in "software mode"

Also Kepler supports DX12 with feature level 11_0 or 11_1 (it depends on the source Nvidia or Microsoft) but certainly not 12_0, but the whole concept of feature levels are completely irrelevant with DX12 core features.

The DX12 "full" support is something that interests gamers or game developers who want to support every feature of the graphics engine of the new API that would make games look more beautiful.

By the way there is no card out there yet that supports every feature of DX12.

[huhn]

directx 12 is slitted in software feature and hardware feature.
if it is important for video processing/decoding/rendering or not we will see in the feature.

the main feature is the possibility to "directly" use GPU resources. which can be called software directx 12.

and my comment about kepler is that they are only feature level 11.0 not even 11.1.

so where i'm now completely wrong and misleading?

think about what you say sometimes...

[NikosD]

Advices to the others...

The world would definitely be a lot better if we 've been following every advice we give to the others...

[P.J]

Can DX12/WDDM 2.0 do anything with HEVC/10bit decoding? :confused:

[huhn]

never heard of an directcompute decoder/hybrid decoder and ASIC decoders doesn't care anyway. the API is still DXVA 2/DXVA

and i don't know if directcompute benefits from dx12 at all.

could be very useful for rendering and presentation.

[NikosD]

looks like MS MF HEVC decoding with windows 10240 is broken. did someone have the same problem?


Anyway, regarding MS MFT HEVC decoder, it didn't work.


I decided to put Win 10 to my main system (signature) and MS MFT HEVC decoder works just fine.

I don't know why MS MFT HEVC using the same OS version doesn't work with my backup system.

[huhn]

works fine on my nvidia system too.

there where some widows updates have to check my AMD system again.

[huhn]

i haven't installed the updates yet but it is now working. what ever it was it's gone.

MS MF HEVC is still at 59 FPS for my i3 4120. so microsoft hasn't touch this decoder. it is still way faster than lavfilter x64 so well done MS.

[NikosD]

i haven't installed the updates yet but it is now working. what ever it was it's gone.

I finished my upgrades of 4 systems (1 laptop and 3 desktops) and only one - my primary signature system - has a working MS MFT HEVC decoder!

For all the other 3 is not working, at least using DXVA Checker.

But even in the signature working system, one very weird thing happened.

When I try an .mp4 HEVC file, the MS decoder doesn't appear.

Then I try a different .mkv HEVC file and it appears.
And then, when I go back to the first not working .mp4 file, HEVC decoder suddenly works for that too!

It's a bug/ conflict of DXVA Checker or OS.

@huhn

Do you have a yellow question mark in device manager for the Intel HDMI audio driver with a Code 10 error ?

I think that Intel driver didn't install correctly the audio driver or I urged a restart while it was installing (I don't remember exactly)

My last weird thing regarding Win 10 installation, is that the new tab functionality while using a Web browser by pressing the mouse wheel, doesn't work with my wireless RF mouse, but it works with my wired USB mouse on the same laptop!

Oh God of Microsoft!

What's going on ?

[huhn]

i have no yellow question marks in my device manager but i'm not actively using an intel GPU they are just installed.

i tried an mp4 HEVC file and it worked frist try. of cause i didn't loaded any other videos in DXVA checker first.

[NikosD]

if windows 10 gets window mode with 10 bit i wouldn't be shocked a direct x 12.1 card is needed.

After the installation of the new Intel drivers for Win 10 v4256, the HDMI device disappeared completely, so now I have no yellow marks :)

The most important thing though, is that DXVA Checker reports P010 output for 10bit H.265 files during playback with MS MFT HEVC decoder...in window mode !!

[huhn]

After the installation of the new Intel drivers for Win 10 v4256, the HDMI device disappeared completely, so now I have no yellow marks :)

The most important thing though, is that DXVA Checker reports P010 output for 10bit H.265 files during playback with MS MFT HEVC decoder...in window mode !!

that doesn't mean a renderer has outputted 10 bit at all.

[NikosD]

so my system isn't broken good to know.

I finished my upgrades of 4 systems (1 laptop and 3 desktops) and only one - my primary signature system - has a working MS MFT HEVC decoder!

For all the other 3 is not working, at least using DXVA Checker.


I finished installing Win 10 to all my 5 systems (4 desktops 1 laptop) and I found out a pattern, that seems a little strange about when MS MFT HEVC decoder works and when it doesn't.

Well, only one system enables MS MFT HEVC decoder and that is the system with a WDDM 2.0 driver (Haswell) system.

On all my other systems with a WDDM 1.3 driver, there is no MS MFT HEVC decoder enumerating in DXVA Checker or used in WMP12 (MFT player), although it is definitely installed in the system - I can see it in MFT Viewer.

Is this really true ?

Does Microsoft enable its MS MFT HEVC decoder only on systems with WDDM v2.0 GPU drivers ?

More feedback is needed by any of you who have installed Win 10 Build 10240.

[06_taro]

On one of my old laptop with only WDDM 1.1 support, MS MFC HEVC decoder is installed but not usable. Haven't try on other PCs as I'd like to receive official upgrade for them.

[NikosD]

So you' ve just confirmed me.
Thanks.

[sneaker_ger]

Have you tried just playing in Windows Media Player? Maybe it's only a quirk of DXVA checker.

[NikosD]

Have you tried just playing in Windows Media Player? Maybe it's only a quirk of DXVA checker.
From the post above:

"On all my other systems with a WDDM 1.3 driver, there is no MS MFT HEVC decoder enumerating in DXVA Checker or used in WMP12 (MFT player), although it is definitely installed in the system - I can see it in MFT Viewer."

[nussman]

So no HEVC playback in WMP?

[NikosD]

No MS MF HEVC decoder in WMP with non WDDM 2.0 GPU drivers.

But WMP can use Lentoid or LAV HEVC decoders for systems with older than WDDM 2.0 GPU drivers.

[NikosD]

DX12 for Fermi will come later

http://nvidia.custhelp.com/app/answers/detail/a_id/3707/

[P.J]

Yea, WMP couldn't play HEVC without breaking the CPU, much worse than LAV CPU
DXVAChecker can decode with DXVA but horrible playback and breaks the CPU

[clsid]

It is strange that software decoding also doesn't work. That shouldn't depend on the driver. Oh well, I prefer LAV anyway. At least it properly falls back to using DirectShow without needing tweaks.

[NikosD]

I don't think that MS MFT HEVC decoder has GPU accelerated capabilities yet.

I don't know if it works with DXVA Nvidia pure HW HEVC decoder, but last time I tested that with Intel Hybrid decoder it didn't work.

So, I was referring to SW MS HEVC decoder not working with older than WDDM 2.0 GPU drivers.
It sounds weird, but probably MS wants to be sure that the GPU driver can output 10bit format properly regardless of the type of decoding (SW/HW).

[huhn]

but you don't need WDDM 2.0 to output 10 bit even windows 7 can do 10 bit output

[NikosD]

In window mode ?

[huhn]

In window mode ?

that was rumoured for windows 10 do you have a conformations?

10 bit fullscreen exclusive is from directx feature level 10.x. so you can already assume you need a new feature level that can do this and so most likely not yet existing hardware if they even add 10 bit window mode.

[Motenai Yoda]

I was on an insider build and IIRC ms's hevc decoder plays well even if my gpu is an old low_end one.
with the upgrade the app plays only audio and wmp refuse or use lav

[NikosD]

Yes, it seems that MS surprised us all and decided just the last minute before the official release to limit the HEVC decoder only to WDDM 2.0 GPU hardware.

What a nice surprise !

[sneaker_ger]

Is the limitation really for WDDM 2.0 or for GPUs with HEVC DXVA support? Maybe they got scared by HEVC Advance.

[NikosD]

Interesting story about HEVC advance...It's all about money (what a surprise)

[Aleksoid1978]

If who interest - Win 10 + GTX 960, vanilla EVR accept P010 for HEVC DXVA 10 bit :) Test in MPC-BE.

[NikosD]

My Haswell's iGPU accepts P010 using MediaFoundation EVR and MS HEVC decoder for 10bit HEVC.

I guess GTX 960 accepts that too and adds to the equation DirectShow EVR with DS HEVC decoder for 10bit HEVC even using DXVA.

[CruNcher]

My guess is security reasons (Content Protection) ;)
WDDM 2.0 comes not only with advances in Performance and such but it needs the Hardware to securely transmit things more efficiently combined with Uefi and SecureBoot and DXVA Protected Mode ;)
Lot of Layers of Security are enabled now ;)

[NikosD]

OK...This is a useful post, I think, about one of the most weird decoders I have ever seen - MS MFT HEVC decoder of Win 10.

After a few tests and a few thoughts about this decoder, here is the situation.

MS MFT HEVC decoder is not a DXVA2 decoder.

It's a D3D11 video accelerated decoder only, which means a few things.

D3D11 video acceleration is something that some developers are claiming that you could use in Win 7 with Platform Update installed and create devices with feature level 9_3 in order to work with it, but for example MediaSDK from Intel couldn’t work even in that way because it needs Direct3D 11.1 which is available in Win 8 or above.

So, according to my findings, D3D11 video acceleration is mainly a Win 8 or above feature, although it could work with Win 7 + Platform Update.

Regarding GPU companies and their drivers, it seems that for Intel for example they already have done their job.

MPEG2_VLD,
H264_VLD_NoFGT,
VC1_VLD2000 and
HEVC_VLD_Main

are all DXVA2/D3D11 capable according to DXVA Checker.

So the device decoders are ready to support both modes of HW decode acceleration – DXVA2 and “DXVA11”

Also DXVA11 is similar to DXVA2, the only thing different is the interface above.
Maybe it’s a little trickier too, in order to work compared to DXVA2.

What all these mean ?

It's impossible to connect to EVR with a DXVA11 decoder, like the MS MFT HEVC decoder, so the DXVA Checker could never become red and could never use that decoder to HW accelerate HEVC content because it uses EVR renderer.

So, DXVA Checker will always fall back to SW using MS MFT HEVC decoder in "Playback mode" or "Playback Benchmark" mode as it uses EVR.

BUT using Decode Benchmark mode I managed to HW accelerate HEVC content (from 1080p up to 2160p) in Hybrid mode using GPU ~100% and my Intel Core i7-4790 system using Win 10 x64 and HD 4600 - drivers 4256 (!)

So, it seems that MS MFT HEVC has already enabled HW HEVC acceleration but with two major problems:

a) It's a MediaFoundation only decoder

b) It's a DXVA11 only decoder.

That combination changes the MS MFT HEVC decoder HW acceleration capability, from useful to useless because the two MediaFoundation players that I know - DXVA Checker and WMP - both use EVR renderer so they can't decode and display HEVC content using DXVA11-only decoders, like MS MFT HEVC decoder (!!)

MS MFT HEVC has enabled HW DXVA acceleration that no one can use it.
Only SW fall back for that decoder.

One last thing...

It seems that a system with a Radeon 7750 card (GCN 1.0) and Catalyst 15.7.1 (WDDM 2.0) driver can't use MS MFT HEVC decoder, not even in SW, using Win 10.

It doesn't enumerate in DXVA Checker and it's absent in WMP.

So, my initial thought about WDDM 2.0 drivers seems to be wrong.

Next thing that I thought and asked him if he has an AVX2 capable CPU.

I couldn't think of anything else...

[nussman]

Sorry but I have no idea what you are talking about ...

"DXVA11" = ? MSDN says it is DXVA2, why you think it is "DXVA11"?
What Intel or AMD GPU support HEVC DXVA2?

[NikosD]

As I wrote above, I call D3D11 video acceleration as DXVA11 because it has no name by Microsoft.

DXVA11 = D3D11 Video acceleration

Intel supports both DXVA2/"DXVA11"

I think the others too.

Read more about D3D11 video acceleration in order to understand the differences with DXVA2

[wanezhiling]

http://i3.tietuku.com/a251913f94a9c7c5.png

Congratulations, PotPlayer dev also verified your supposition. :)

[JohnLai]

It seems that Ivybridge still uses 15.33 driver and not 15.36, so I don't expect Ivy to support DXVA HEVC acceleration.

But have you tried Lentoid HEVC OpenCL decoder or Cyberlink's PowerDVD HEVC OpenCL decoder with Ivy to see of there is any HW acceleration from there ?

Because Ivy supports OpenCL.

.....about Ivy Bridge HD4000 OpenCL HEVC decoding using Lentoid OpenCL, I never got it to work when I was using Win7. I can't recall what error pop up last time whenever lentoid is being loaded.
As for cyberlink = I am too poor to afford one.....

Just now, I tried to install lentoid opencl on win10 (free upgrade from win7). Apparently.....there isnt any opencl support for intel hd4000 for win10 as shown by gpuz 0.8.5

[NikosD]

Congratulations, PotPlayer dev also verified your supposition. :)

No, my assumption is wrong because as I wrote in my previous post a Radeon 7750 with Catalyst 15.7.1 and WDDM 2.0, still doesn't have MS MFT HEVC decoder with Win 10.

You have to check my next assumption which is AVX2 instructions.

.....about Ivy Bridge HD4000 OpenCL HEVC decoding using Lentoid OpenCL, I never got it to work when I was using Win7. I can't recall what error pop up last time whenever lentoid is being loaded.
As for cyberlink = I am too poor to afford one.....

Just now, I tried to install lentoid opencl on win10 (free upgrade from win7). Apparently.....there isnt any opencl support for intel hd4000 for win10 as shown by gpuz 0.8.5

Cyberlink's PDVD 15 HEVC decoder is free, because you can take it from the trial PowerDVD 15.
I'll give you instructions how to get it.

Also, GPU-Z is not so reliable for new systems (HW or SW)

I would definitely try realworld OpenCL tests like LuxMark or GPU Caps Viewer which has some built-in easy to test OpenCL tests.

https://forum.beyond3d.com/threads/direct3d-feature-levels-discussion.56575/page-27#post-1865163

Skylake Gen9 GPU supports Feature Level 12_1, it's the most advanced FL12_1 GPU for integrated graphics, AMD can't support FL12_1.

I've read so far only Anandtech's review and I'm shocked by the size of the failure of new Skylake CPU.

For example:
The Core i7-6700K (Skylake) is 5% faster on average than Core i7-4790K (Haswell)

For two generations (Skylake to Haswell) this is simply unacceptable.

Also Skylake is slower in IPC than previous generation Broadwell. This is simply awful.

Also I copy a phrase from Anandtech's review



Conclusions on Gaming

There’s no easy way to write this.

Discrete graphics card performance decreases on Skylake over Haswell.



Unbelievable!

Also, the IGP of Core i7-6700K is about 20-25% on average faster than Core i7-4770K and a lot slower than Core i7-5775C in games, according to Anandtech.

No HDMI 2.0

Besides HW HEVC decoding (8bit only) and encoding, I would call Skylake a big failure.

[Plutotype]

I have also read the reviews about Skylake. I also havent seen any slide or source mentioning Skylake desktop CPU having GT4e iGPU ( Iris Pro HD580 ). That should be compared to GT3e Iris Pro HD6200 on i7-5775C, btw, not the GT2 on i7-6700K.
In terms of HDMI2.0, is this not a motherboard thing? Some Z170´s have it definately. Why would a Z170 motherboard have HDMI2.0 port when the iGPU on the Z170 CPU cant support it?

Regarding 10-bit HEVC content ( upcoming UltraHD blu-ray ), will Skylake iGPUs ever support full hardware decoding of such content?
If no, there is no need upgrade from Broadwell, which at least has these i7/i5 models. Now its clear why they put these two models out.

Can i7-5775C hybrid decoder decode 60fps high 50Mbps 5.1 level 10-bit HEVC content and not go beyond 20-30% of CPU load?

[huhn]

You have to check my next assumption which is AVX2 instructions.

my i7 3770k doesn't have AVX2 but the decoder was working with my gtx 760. now i put my r9 270 in the system because the gtx 760 just died and now the decoder doesn't work at least not now.

but AVX2 can't be the reason.

[NikosD]

It's not WDDM 2.0 (because AMD cards with WDDM 2.0 do not work) and it's not AVX2 because Ivy works.

OK...

So, it needs WDDM 2.0 but not AMD cards...Simple :)

[huhn]

my HTPC with AVX2 and the same card had a working decoder.

[NikosD]

Someone with a lot of different generations HW can do the combinations in order to find out what's going on before the official explanation - I hope to give that explanation - from MS.

Looks like something with WDDM 2.0 and AVX2.

[JohnLai]

NikosD, by removing my discrete gpu and connecting my display using only IGPU HD4000, gpuz finally shows opencl checkbox is checked.

~.~ Geez, turns out the dll/ax is 32bit. I had to download x86 MPC-HC to test.

Anyway...I attached the results here.
I dont see much different in term of cpu and igpu load when testing both HAM and SW option.

[NikosD]

I can't see yet the images, but it's better to do some tests with DXVA Checker in order to see some figures and CPU/GPU usage during HAM and SW tests.

In "playback benchmark" mode or even better in "decode benchmark" mode (to isolate the GPU usage for other reasons than HEVC decoding)

[Yups]

Also Skylake is slower in IPC than previous generation Broadwell. This is simply awful.



You should make it clear you compare it to Broadwell-K with 128 MB of edram. This is the only reason why certain benchmarks have a better IPC than Skylake without L4. Non edram BDW won't have better IPC.



In terms of HDMI2.0, is this not a motherboard thing? Some Z170´s have it definately. Why would a Z170 motherboard have HDMI2.0 port when the iGPU on the Z170 CPU cant support it?


HDMI 2.0 on Skylake is working over Thunderbolt. Boards with Alpin Ridge support HDMI 2.0.

Btw,

http://www.hardware.fr/articles/940-23/hd-graphics-530-pratique-h-265-consommation.html


Does Quicksync HEVC run out of the box?

[NikosD]

You should make it clear you compare it to Broadwell-K with 128 MB of edram. This is the only reason why certain benchmarks have a better IPC than Skylake without L4. Non edram BDW won't have better IPC.


It wasn't necessary to say that, because here we compare desktop products, Skylake isn't available for other platforms yet.

All 5 different SKUs of desktop Broadwell have the eDRAM.

You can't buy a desktop Broadwell without one.



Btw,

http://www.hardware.fr/articles/940-23/hd-graphics-530-pratique-h-265-consommation.html


Does Quicksync HEVC run out of the box?

Yes, apparently it does.

And why not ?

DXVA decoding is a standard and if the DXVA device decoders installed by the drivers of the GPUs and the developers of the DXVA decoders follow exactly Microsoft's standards, you don't have to explicitly make changes for a new platform.

Of course some kind of optimizations always occur, depending on the GPU.

[Plutotype]

But Skylake does not offer 10-bit HEVC full hardware decoding, am I right? Only 8-bit.
Thanks

[wanezhiling]

But Skylake does not offer 10-bit HEVC full hardware decoding, am I right? Only 8-bit.
Yes..
And this is the worst thing of the new generation, I think..:mad::mad::mad:

[Aleksoid1978]

About MS MFT HEVC decode in Win 10.
I have 2 PC with Win 10 Ent x64. All activated. Use latest drivers from Nvidia/AMD sites.
One with GTX 960, second with ATI HD 7750. On both dxdiag show WDM 2.0

In DXVAChecker i see Microsoft H265 Video Decoder MFT decode in DSF/MFT Viewer->Media Foundation. But - when try Check Decoder -> Media Foundation - there is no MS MFT HEVC decode ((

P.S. Microsoft H264 Video Decoder MFT working normal.

[NikosD]

I have 2 PC with Win 10 Ent x64. All activated. Use latest drivers from Nvidia/AMD sites.



What CPUs ?


One with GTX 960, second with ATI HD 7750. On both dxdiag show WDM 2.0



That is new.

You are saying that even with GTX 960 you don't see MS MFT HEVC decoder in DXVA Checker.

I'm starting to think that it's simply a MS bug eventually.

[Aleksoid1978]

I7 2600K + GTX 960
I7 930 + ATI 7750.

On both PC i see MS H265 MFT Decoder in DXVA Checker->MFT Viewer but can't use not for Decode not for Playback ... i think too that it's a some bug.

[NikosD]

No AVX2 for you.

If you manage to find a Haswell processor, you will probably see the MS MFT HEVC decoder enabled.

[Aleksoid1978]

No AVX2 for you.

If you manage to find a Haswell processor, you will probably see the MS MFT HEVC decoder enabled.

Where are rolling this world - for the video decoder is required almost the most modern processors :)

I'd rather use DXVA2 + GTX 960 :)

[NikosD]

Undoubtedly the best "review" of desktop Skylake which "presented" - so to speak - yesterday.

http://semiaccurate.com/2015/08/05/intel-plays-press-skylake-stupidity/

[Yups]

It wasn't necessary to say that, because here we compare desktop products, Skylake isn't available for other platforms yet.


Broadwell isn't desktop only. And you should be clever enough that edram vs no edram isn't a neck to neck comparison. If you are surprised that an edram version has some advantages over a non edram version, you aren't that clever.

[NikosD]

After all these faulty thoughts that you have made regarding Skylake and the comparison with previous gens of Intel CPUs, I can clearly say that you shouldn't talk about cleverness...

Unless of course you are an Intel seller (that would be an excuse in more than one ways)

Then I would urge you to read the semi - accurate's "review".

You might understand it better than Anandtech's.

[JohnLai]

I can't see yet the images, but it's better to do some tests with DXVA Checker in order to see some figures and CPU/GPU usage during HAM and SW tests.

In "playback benchmark" mode or even better in "decode benchmark" mode (to isolate the GPU usage for other reasons than HEVC decoding)

I cant get PDVD video decoder or PDVD HAM decoder to show at dxvachecker after I try to load the hevc video using 'Check Decoders..' (tried both 64bit and 32bit).
Only 'DS32 LAV Video Decoder [HVC1 3840x2160]' shows 'Unsupported' and '[MF32] Microsoft H265 Video Decoder MFT [D3D11] [HVC1 3840x2160]' shows 'HEVC_VLD_Main' + 'HEVC_VLD_Main10'.


Now I think about it, isn't the EVR statistic decoder output on my second screenshot of 'HAM' selection should show 'Cyberlink HAM Decoder (PDVD Generic)' instead of 'Cyberlink Video Decoder (PDVD Generic)' ?

I only have to run 'regsvr32.exe CLCvd.ax' as administrators, no?

[NikosD]

As I wrote you in PM you have to increase the merit to normal or more and choose 32 bit because PDVD is 32 bit only.

[JohnLai]

As I wrote you in PM you have to increase the merit to normal or more and choose 32 bit because PDVD is 32 bit only.

Right after I load the Elecard_4K_video.h265 when selecting 32bit at dxva checker.


http://imgur.com/7pHUSTX

[NikosD]

Find a .mkv or .mp4 file, check the merit of Cyberlink DXVA filter (not the HAM filter) that is normal or upper, select from the properties of the cyber link filter inside DXVA Checker "Enable HAM", NOT the Cyber link (HAM) filter, the other filter.

[huhn]

Where are rolling this world - for the video decoder is required almost the most modern processors :)

I'd rather use DXVA2 + GTX 960 :)

I was using the decoder on a none AVX2 system so that's not the reason. there is something buggy not sure what. after the release of 10240 I cannot use the decoder on any of my PC it simply didn't worked.

[JohnLai]

Find a .mkv or .mp4 file, check the merit of Cyberlink DXVA filter (not the HAM filter) that is normal or upper, select from the properties of the cyber link filter inside DXVA Checker "Enable HAM", NOT the Cyber link (HAM) filter, the other filter.

http://i.imgur.com/aE035Q7.jpg?1

Finally got it to show after remuxing the Elecard_4K_video.h265 using mkvmerge GUI from mkvtoolnix.

Merely select 'Enable HAM'?

How about the 'Property...' near the DXVA?

[NikosD]

"Enable HAM" from DXVA button of DXVA Checker.
The property page of filter works only during playback, IIRC

[JohnLai]

"Enable HAM" from DXVA button of DXVA Checker.
The property page of filter works only during playback, IIRC

The result after select 'Enable HAM'.

http://i.imgur.com/b2CM6zl.jpg?1

No GPU usage at all aside from temperature increase.

[NikosD]

Is OpenCL active ?
Like when you remove your discrete card.

Because it looks like it's in SW fall back mode.
The CPU usage is very high, the GPU usage very low.

[JohnLai]

Is OpenCL active ?
Like when you remove your discrete card.

Because it looks like it's in SW fall back mode.
The CPU usage is very high, the GPU usage very low.

http://i.imgur.com/tHWPvrV.gif?1

Note: I downgraded the driver to 4226 instead of default win10 4252 because of the IGPU low clock issue.

EDIT: Just in case, I let win10 autoupdate to 4252 to see if there will be any different with dxvachecker result. There is no difference in the result. Exactly the same high cpu usage with no gpu usage.

[NikosD]

According to the official reply of MS here

http://answers.microsoft.com/en-us/windows/forum/windows_10-windows_install/has-microsoft-playready-3-broken-windows-10-hevc/8a17d615-5d0d-410d-8d18-6e951f108d11?auth=1

only systems with HW HEVC acceleration will get the SW HEVC decoder of MS!

Check your system with DXVA Checker.

If you have a HEVC_VLD_Main device decoder, then you will get the MS MFT HEVC decoder in Win 10.

If you don't...Bad luck!

[Aleksoid1978]

Read this : "The list of supported codecs for Windows 10 can be found here: https://msdn.microsoft.com/en-us/library/windows/apps/xaml/mt282148.aspx."
And i see that H265 support only for Codec/Container - fMP4 and MPEG-4
I check my PC with GTX 960 again, and that's right - works with MP4/MKV, don't work with MPEG-TS.
Tested - and yes MS MFT H265 Video Decode use GPU decoding(D3D11), i test on 4k and have 0-1% CPU.

Later test on 4k 10bit and upload result, compare with MPC-BE's video decoder :)

[Aleksoid1978]

Decode benchmark result:

Decoder: Microsoft H265 Video Decoder MFT
[1]
Decoder Device: HEVC_VLD_Main10
Frames: 5321
Time: 40,826 sec
FPS: 130,333 [119-145] fps
CPU Usage: 0 [0-0] %
GPU Usage: -
[2]
Decoder Device: HEVC_VLD_Main10
Frames: 5321
Time: 40,466 sec
FPS: 131,492 [121-149] fps
CPU Usage: 0 [0-2] %
GPU Usage: -
[3]
Decoder Device: HEVC_VLD_Main10
Frames: 5321
Time: 40,453 sec
FPS: 131,534 [120-149] fps
CPU Usage: 0 [0-1] %
GPU Usage: -


Decoder: MPC Video Decoder
[1]
Decoder Device: HEVC_VLD_Main10
Frames: 5321
Time: 40,812 sec
FPS: 130,378 [121-146] fps
CPU Usage: 0 [0-2] %
GPU Usage: -
[2]
Decoder Device: HEVC_VLD_Main10
Frames: 5321
Time: 40,513 sec
FPS: 131,342 [121-149] fps
CPU Usage: 0 [0-2] %
GPU Usage: -
[3]
Decoder Device: HEVC_VLD_Main10
Frames: 5321
Time: 40,487 sec
FPS: 131,424 [120-149] fps
CPU Usage: 0 [0-1] %
GPU Usage: -


for bench i made short mkv from "Astra Ultra HD Demo.ts". Here is sample (https://yadi.sk/d/V0bpah3riLMHr)

MediaInfo:

Video
ID : 1
Format : HEVC
Format/Info : High Efficiency Video Coding
Format profile : Main 10@L5.1@Main
Codec ID : V_MPEGH/ISO/HEVC
Duration : 1mn 46s
Width : 3 840 pixels
Height : 2 160 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 50.000 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 10 bits
Default : Yes
Forced : No


What interesting - MS MFT H265 Decoder can't playback with DXVA, only in software but P010 output with normal picture. MPC-BE's decoder in software can output P010 too, but green picture :)

[NikosD]

Looks like your decoder has exactly the same speed with MS MFT HEVC in HW.

That would probably mean that even when you and MS, leverage different framework (DirectShow vs MediaFoundation) and different API (DXVA2 vs D3D11) you both manage to utilise the HEVC fixed function unit at 100%


What interesting - MS MFT H265 Decoder can't playback with DXVA, only in software


MS MFT HEVC decoder is D3D11 only - No DXVA2 as I wrote in a previous post.
And because D3D11 is incompatible with EVR, it falls back to SW decode.

"Decode benchmark" is renderless and can use D3D11 for HW acceleration.

but P010 output with normal picture. MPC-BE's decoder in software can output P010 too, but green picture :)

The strange think here for 10bit HEVC is that with my system which doesn't have HEVC_VLD_Main10, I can output P010 in SW decoding of normal playback, but in "Decode benchmark" uses NV12 (!)

[mandarinka]

Edit:

AMD A10-8700P (Carrizo), seems it does 4K, but no Main 10.

http://www.notebookcheck.com/fileadmin/Notebooks/HP/Pavilion_17-g054ng/dxva_carrizo.png

http://www.notebookcheck.com/Test-HP-Pavilion-17-g054ng-Notebook.147334.0.html

[NikosD]

I don't know why all those hardware review sites with all those benchmark numbers don't get a little bit in the trouble to measure the actual performance in terms of raw speed and maybe power of those fixed function units.

Just as they test CPU, 3D, Games etc, they could probably add video acceleration benchmarks too.

I can't buy all the hardware in the world to test HEVC hardware :)

If someone sends it, I would do it.

Right now there are 4 HW fixed function units, claiming pure HEVC HW acceleration.

First it was Nvidia wkth 960 and then Intel's Atom, if I remember the product exactly.

Then we have Carizzo and now Skylake.

It would be an interesting comparison under the new Win 10 OS.

[littleD]

Not quite benchmark, but still some test http://www.hardware.fr/articles/940-23/hd-graphics-530-pratique-h-265-consommation.html

[Stereodude]

Has anyone confirmed that the various HEVC decoders actually are decoding all 10-bits of 10-bit HEVC? I've seen some of the Kodi/OpenELEC developers say that Nvidia's 10-bit HEVC decoder in the 960 truncates 10bit HEVC to 8 bits and decodes that (ie: doesn't decode all 10 bits).

[NikosD]

It doesn't sound too easy task.

How could someone do that ?

I don't have a HW 10bit decoder to test it (pure or hybrid) but I could test SW 10bit decoders.

[Stereodude]

It doesn't sound too easy task.

How could someone do that ?
Are you referring to the task of verifying it decodes all 10 bits or the task of dropping the 2 least significant bits during the decode (or just ahead of it)?

[NikosD]

About verification of 10bit decoding.

[Stereodude]

Wouldn't it just be a matter of decoding a 10-bit gradient and comparing that to the original gradient source?

Here's where I saw one of the Kodi developers say it doesn't really do 10bit HW decoding (http://forum.kodi.tv/showthread.php?tid=228464&pid=2072677#pid2072677).

[NikosD]

I think the same or similar trick did AMD with 10bit HW decoding of H.264.

[sneaker_ger]

Has anyone confirmed that the various HEVC decoders actually are decoding all 10-bits of 10-bit HEVC? I've seen some of the Kodi/OpenELEC developers say that Nvidia's 10-bit HEVC decoder in the 960 truncates 10bit HEVC to 8 bits and decodes that (ie: doesn't decode all 10 bits).
I doubt that. At least in AVC you will get horrible artifacts in some scenes and we have seen no reports of that happening. It should be fairly simple to test since HEVC decoding is deterministic. Use LAV copy-back and software and compare (even if dithered to 8bit by LAV in your chain). I don't have a GTX 960 so I cannot do it myself.

[nevcairiel]

DXVA2 should be proper full 10-bit decoding and unmolested output. Its possible that linux decoding APIs don't support full 10-bit output.
But if it makes all of you happy, I can verify that a bit later.

[NikosD]

Use LAV copy-back and software and compare (even if dithered to 8bit by LAV in your chain). I don't have a GTX 960 so I cannot do it myself.

How can be sure that SW decoding is 10bit and how can be sure that after decoding will get 10bit output ?

Are you sure that you can actually see differences between 10bit and 10bit dithered ?

I'm not sure that my monitor even supports 10bit.

[sneaker_ger]

I didn't say you should use your eyes for comparison. Use something proper like a checksum over all pixels.

[NikosD]

DXVA2 should be proper full 10-bit decoding and unmolested output. Its possible that linux decoding APIs don't support full 10-bit output.
But if it makes all of you happy, I can verify that a bit later.
That would be the best, because we are not talking only about API support but HW support too.

And tricky drivers of course, if they are indeed tricky.

[nevcairiel]

I just did a quick check on two 10-bit clips, and after skipping the first couple frames which had corruption (because Software uses grey to fill corrupted areas and hardware uses black), DXVA2-CB on the GTX960 and Software decoding are identical.
So I think its safe to assume that DXVA2 on the GTX960 at least delivers proper unmolested 10-bit content.

[Stereodude]

Okay thanks for the confirmation.

[P.J]

I think the same or similar trick did AMD with 10bit HW decoding of H.264.

Can AMD play H.264 10bit? :confused:

[nevcairiel]

Can AMD play H.264 10bit? :confused:

No, but some people claimed they can by simply running it through the 8-bit decoder, which had crazy image artifacts in most clips.

[Yups]

Any demanding HEVC videos I could test on my new i7-6700k?

DXVA screenshot (driver 10.18.15.4256 Windows 10)

http://fs1.directupload.net/images/150815/kw48d4fu.png

[JohnLai]

Any demanding HEVC videos I could test on my new i7-6700k?

DXVA screenshot (driver 10.18.15.4256 Windows 10)

http://fs1.directupload.net/images/150815/kw48d4fu.png

How about testing this video?

I have been trying to play this HEVC/h.265 clip (se download link) but it does not play smoothly with my Intel Core i7-4790K iGPU HD4600 graphic on Dell 4k UP3214Q monitor, I get the best result with the MPC-HC player, but it stutters and CPU load is 80-90% can anyone download this clip and try it out?

http://demo-uhd3d.com/fiche.php?cat=uhd&id=42

[NikosD]

Any demanding HEVC videos I could test on my new i7-6700k?


A really quick buyer.

I would start with 8bit HEVC (a few 1080p and more 4K) in order to compare it directly with the other popular HW decoder of Nvidia 960.

Then I would try some 10bit clips too in order to test LAV DXVA native HW decoder with Intel HW (hybrid 10bit decoding) and EVR of Win 10.

Don't expect to decode 10bit 4K HEVC with Intel HD 530 though.

Let the benchmarks begin!

Who has the most powerful fixed-function HW 8bit HEVC decoder, Nvidia or Intel ?

[Yups]

How about testing this video?



Very low CPU load with DXVA2 native in MPC-HC. Here a snapshot

http://fs2.directupload.net/images/150815/kir4po6w.png


~2% CPU load. GPUz doesn't fully support Skylake graphics, temperature and GPU clock is wrong. No idea if GPU load is correct. GPU power might be correct because 20W in Furmark seems plausible.

Up to 60% CPU load with no hardware decoder in use. I can play these UHD videos only in 1920x1080 without 4k monitor by the way.

[NikosD]

Very low CPU load with DXVA2 native in MPC-HC.

Right below there is a link of a smaller in duration and file size sample in .mp4, more suitable for benchmarking.

Sample: (4K HEVC 8bit - 25Mbps - 30fps)
http://demo-uhd3d.com/fiche.php?cat=uhd&id=101

Use DXVA Checker in both modes "Decode Benchmark" and "Playback Benchmark" in 1280x720 window scaling with LAV filters 0.65.47 in DXVA native and show us the results. (select a repeat count of at least 3)

It would be interesting to see the SW decoding benchmarks of LAV x64 and maybe Lentoid or MS MFT too, in order to test the new Skylake architecture against previous architectures in SW decoding.

[JohnLai]

Very low CPU load with DXVA2 native in MPC-HC. Here a snapshot

http://fs2.directupload.net/images/150815/kir4po6w.png


~2% CPU load. GPUz doesn't fully support Skylake graphics, temperature and GPU clock is wrong. No idea if GPU load is correct. GPU power might be correct because 20W in Furmark seems plausible.

Up to 60% CPU load with no hardware decoder in use. I can play these UHD videos only in 1920x1080 without 4k monitor by the way.

Interesting.....cause even core i3-4330 HD4600 and GTX 970 hybrid DXVA cant play the video smoothly.

[Yups]

Right below there is a link of a smaller in duration and file size sample in .mp4, more suitable for benchmarking.

Sample: (4K HEVC 8bit - 25Mbps - 30fps)
http://demo-uhd3d.com/fiche.php?cat=uhd&id=101




Very low CPU load with this video, 1-3%. CPU clock stays low too.

[NikosD]

Without DXVA Checker's results, "very low CPU load" means nothing really useful.

Because a HW VLD decoder always off - loads CPU keeping CPU usage <5%

The main aspect is how fast is that HW decoder.

[nevcairiel]

Interesting.....cause even core i3-4330 HD4600 and GTX 970 hybrid DXVA cant play the video smoothly.

Its an 8-bit clip, and Skylake has a full fixed function 8-bit hardware decoder, if it couldn't at least decode in realtime for playback, that would've been silly. ;)

[JohnLai]

Its an 8-bit clip, and Skylake has a full fixed function 8-bit hardware decoder, if it couldn't at least decode in realtime for playback, that would've been silly. ;)

Speaking of fixed function hardware decoder......

Is it possible to use DXVA (Copy Back) on secondary gpu? Such as providing a selection of gpu for DXVA copy back?

The scenario, if I install GTX 950 (once it is released later) at my second PCIe x16 slot on my system with GTX970 as primary gpu, I would like to use GTX 950 for DXVA decoding instead of GTX 970. (with Intel IGPU enabled too)
Thus, there will be 3 different GPUs.


Apparently, selecting Intel Quicksync as hardware decoder works with gtx 970 as primary gpu.

Intel IGPU(Quicksync) + nvidia gpu(vp6) = it works.

nvidia gpu (VP6)(Primary gpu for display) + nvidia gpu (VP7)(Secondary GPU) = possible with dxva copy back?

[nevcairiel]

Its technically possible, but only if a screen is connected.
A future version of LAV may have a device selection

[JohnLai]

Its technically possible, but only if a screen is connected.
A future version of LAV may have a device selection

I thought from Windows 8 onwards, users no longer need to use the 'fake display' like the one described here http://www.helping-squad.com/fake-connect-a-monitor/ to make use of active gpu without any screen being connected?

Win8 has support for headless display mode as stated by microsoft. As long as compatible wdmm 1.2 drivers is used.

http://download.microsoft.com/download/6/E/5/6E5329B2-F1CC-4922-A74A-EC91267B996D/windows-display-driver-model-enhancements.docx

Page 28?

[nevcairiel]

That only works with DX10/11, not with DX9, which is used for DXVA2 usually.

[JohnLai]

That only works with DX10/11, not with DX9, which is used for DXVA2 usually.

Oh...okay....I understand.
:thanks:

[NikosD]

I thought from Windows 8 onwards, users no longer need to use the 'fake display' like the one described here http://www.helping-squad.com/fake-connect-a-monitor/ to make use of active gpu without any screen being connected?

Win8 has support for headless display mode as stated by microsoft. As long as compatible wdmm 1.2 drivers is used.


You need a D3D11 decoder which doesn't exist using DirectShow.

All existing DirectShow decoders are DXVA2.

QuickSync decoder by Eric Gur implemented in LAV filters and PotPlayer, does support D3D11 and headless decoding using Win 8 or better because it is based on Intel's MSDK which has added D3D11 support.

[sheppaul]

It is already possible to use iGPU's dxva copyback decoder(dx11) while using external GPU in potplayer. I've been using it for several weeks instead of quicksync. Intel frequently breaks quicksync decoder whenever they release a new driver. So I've switched quicksync to iGPU's dxva copyback decoder. It seems more stable. You can try it if it work in other external GPU except for main gpu.

[nevcairiel]

You need a D3D11 decoder which doesn't exist using DirectShow.

All existing DirectShow decoders are DXVA2.

Copy-Back could use D3D11 if it wanted to, because its entirely self-contained and doesn't have to interact with a renderer - and DXVA-Native doesn't allow a GPU choice anyway, because it needs to use the same GPU as the video is being rendered on.

However, D3D11 DXVA decoding is quite annoying to handle, and not sure the benefits would be worth all the effort and extra maintenance.

[Yups]

Without DXVA Checker's results, "very low CPU load" means nothing really useful.



I don't see "Decode Benchmark" and "Playback Benchmark" in DXVA checker.

[NikosD]

For those interested:

I discovered today an Intel MediaFoundation decoder (DXVA2/D3D11) x86/x64 for M-JPEG using the DXVA2 decoder device with the id "91CD2D6E-897B-4FA1-B0D7-51DC88010E0A" and a description as "Intel Hardware M-JPEG decoder MFT"

Using DXVA Checker of course, I managed to measure the performance against LAV x86/x64 (latest 0.65.47) and built-in MS MFT M-JPEG decoder x86/x64 using:

Win 10 x64 - Core i7 4790 - HD4600@1.5GHz - Drivers 4256

Both LAV and MS decoders are single-threaded and SW only.

Test file:
MJPEG clip - 3840 x 2160 (292Mbps@50fps)

Decode Benchmark mode

GPU usage is dead zero, so the Intel DXVA2 MFT decoder is a pure HW decoder (not Hybrid), because Haswell supports MJPEG decoding in HW.


Intel DXVA2 MFT x64/x86 397/400/397 CPU 10%

MS MFT x64 57/59/60 CPU 12%

MS MFT x86 49/52/54 CPU 12%

LAV x64 43/45/46 CPU 13%

LAV x86 42/44/45 CPU 13%


I found out also Intel's MFT HW VP8 and VP9 decoders using the id "49761BEC-4B63-4349-A5FF-87FFDF088466" which is now called "VP9and8_VLD_ClearVideo" by the new DXVA Checker v3.6.0, but they seem disabled for Haswell and I didn't manage to enable them.

If I do, I'll report back.

[P.J]

DXVA Checker 3.6.0 is out :)

[Yups]

Gen9 details have been released.


http://abload.de/img/1y0qct.png

http://abload.de/img/233p4j.png

http://abload.de/img/3fmpet.png

http://abload.de/img/4udo9j.png

http://abload.de/img/5a4q2x.png
https://hubb.blob.core.windows.net/e5888822-986f-45f5-b1d7-08f96e618a7b-published/fe98b14c-347b-49a7-8094-bfe43d16f67c/SPCS003%20-%20SF15_SPCS003_103f.pdf?sv=2014-02-14&sr=c&sig=f%2BgeA8kQj0rC8gJXIn%2Fl5IvDY25fycAN3rsvP4v%2BJmU%3D&se=2015-08-19T17%3A25%3A06Z&sp=rwd

[NikosD]

I've just read in Intel forums that Broadwell already fully supports VP8 in HW (decode and encode using Windows, for Linux only HW decoding is supported and GPU accelerated (hybrid) encoding)

Also Broadwell supports already VP9 GPU accelerated (hybrid) decoding.

Intel has never advertised Haswell as VP8 capable for HW decoding, BUT the driver installs the video accelerator which is disabled.

VP8 accelerator could be used in hybrid mode for Haswell, like HEVC, but Intel decided obviously that VP8 support belongs to Broadwell and above.

[nevcairiel]

Its a shame they didn't build an official DXVA2 specification for VP8/9 decoding, or released their proprietary way to do it (as it clearly goes through some DXVA2 device), oh well. Only being able to use MediaSDK and not implement it using existing DXVA2 decoders is sad.

[NikosD]

Only being able to use MediaSDK and not implement it using existing DXVA2 decoders is sad.

I have asked Intel about that and I'm waiting for a reply.

In the meantime, how can you be sure that they don't support HW acceleration directly using DXVA but only via MSDK ?

Also, do you have any interest in implementing MJPEG HW decoder now that I'm sure that it uses pure HW decoding for all iGPUs from Haswell and newer ?

[NikosD]

This is not a place to advertise cards.

It is different to comment on cards, to compare/ review cards than pure advertising of the OEM's pages.

The links should be deleted.

[P.J]

Waiting for GTX 950's 4K benchmark :)

http://www.computerbase.de/2015-08/nvidia-geforce-gtx-950-test/5/

[NikosD]

It's interesting that those R9 285 AMD cards with the new 4K H.264 decoders are real screamers.

After Anandtech that wrote first that they are faster even than Nvidia's VP6/VP7 (Maxwell 1/2), now it's the second site saying the same thing.

AMD was the last company that brought us 4K H.264, but they are the fastest too (for discrete cards - Intel's QuickSync is unbeatable in pure HW decoding but looses a lot in real playback using a renderer due to low GPU power - still even after that remains the top H.264 decoder even in real playback)

[nevcairiel]

In that particular benchmark, in pure decode, the NVIDIA cards are faster at H.264, the R9 380 probably just has a lot more processing power to actually be faster during playback (including rendering)
Oddly enough, in 1080p, the AMD cards are actually slower in the pure decode benchmark rather than the playback benchmark. I wonder if they fuzzed something up while testing... Maybe the AMD numbers are flipped, between decode and playback? That would make more sense.

[NikosD]

How do you know that ?

The figures of the results say different things.

If you have done tests on your own or read it somewhere else, please repost the results here.

GTX 980 should be faster than R9 380 but still looses in real playback according to the charts.

If you challenge figures publicy available, you must provide proofs.

[nevcairiel]

I just read their charts (the computerbase.de P.J linked), nothing else. Some key numbers:

Playback 1080p, R9 380: 419.2 fps
Playback 1080p, GTX960: 334 fps

Decode 1080p, R9 380: 290.4
Decode 1080p, GTX960: 443.0

That seems like odd numbers to me, and make me question the test. NVIDIA gets faster when doing pure decoding, which is expected, and the FPS number matches what I know from my own tests on a GTX 960.
But AMD gets slower? Thats just weird.

I don't need to provide proof to mention that the numbers are looking weird. Maybe its normal that they act like that because of something special about AMD, but maybe not.
Can't we discuss results without calling for proof or sources for every opinion? :P

[madshi]

I've read somewhere that the very latest AMD GPUs now have hard wired processing for certain DXVA algorithms like chroma upscaling, scaling, color conversion etc (same as Intel), while NVidia probably still uses shaders for that. That would explain why AMD might be faster than NVidia when doing playback, but slower than NVidia when just decoding. However, if it's the same clip those numbers are from then it doesn't make a lot of sense that playback is faster than decoding on R9 380 - unless decoding involves copyback or something...

[NikosD]

@nevcairiel

Yeah..I've just seen the next page with decode results.

It looks like they did something wrong in testing or displaying the results.

But I remember for sure that anandtech had reported R9 285 as faster than Nvidia.

You can discuss anything, I hadn't seen the next pages I read it via my mobile phone with a google translator of Chrome.

I don't speak German.

@madshi

Nice catch If they changed to copy-back while testing the others in native.

That would explain the crazy results.

[NikosD]

I've noticed that most of the people here in Doom9 don't have recent cards like R9 285 or Carrizo or some other interesting AMD cards regarding fixed-function HW or even if they do, they don't know or not interested in testing them in benchmark mode.

I'm afraid the same thing is happening with professional HW review sites - as I've written recently - so there is lack of information regarding AMD fixed-function HW.

I can't buy everything to test it!

[Yups]

i7-6700k HD 530 1.15 Ghz
10.18.15.4274
Windows 10 x64

bbb-3840x2160-cfg02.mkv (Big Buck Bunny 4k 60 fps 8 bit HEVC)
http://www.libde265.org/downloads-videos/


Decode Benchmark (video size was greyed out)
http://fs1.directupload.net/images/150822/mxr26oam.png

Playback Benchmark 1280x720
http://fs1.directupload.net/images/150822/zqqjrbbg.png


Does a Nvidia GTX 970 support 4k HEVC in hardware? I could compare.

[nevcairiel]

Does a Nvidia GTX 970 support 4k HEVC in hardware? I could compare.

No, for NVIDIA its only GTX 960 and the newly released 950. All others do hybrid.

[Yups]

i7-6700k HD 530 1.15 Ghz
10.18.15.4274
Windows 10 x64


LG_4K_View-the-Feeling.mp4 (4k 30 fps 8 bit HEVC)
http://demo-uhd3d.com/fiche.php?cat=uhd&id=101


Decode Benchmark (video size was greyed out)
http://fs2.directupload.net/images/150822/rrsi6imi.png

Playback Benchmark 1280x720
http://fs2.directupload.net/images/150822/trx6glxo.png

[NikosD]

Looks like a winner.

P.J, JohnLai, Nevcairiel or someome else could add his Nvidia 960 result on the same clips.

I think Skylake is faster.

Looking at some previous results of Nvidia 960, I wouldn't call Skylake just faster.

I think it is almost 50% faster than 960, which has a 4K@120 FPS decoder for 10bit clips and Skylake looks like 4K@180 FPS decoder for 8bit clips.

[nevcairiel]

I don't have the 960 setup for benchmarking right now, but Yups could benchmark some of the clips I used back in January:
http://forum.doom9.org/showthread.php?p=1706902#post1706902

Specifically the high bandwidth Samsung clip would be interesting (SamsungNX1.Leaves.2160p@30fps-80Mbps - direct link (http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0235-UHD_Sample1.MP4))

[P.J]

GTX 960, Windows 10 x64


LG_4K_View-the-Feeling.mp4 (4k 30 fps 8 bit HEVC)
http://demo-uhd3d.com/fiche.php?cat=uhd&id=101


Decode Benchmark (video size was greyed out)
http://i.imgur.com/iTxohMY.png

Playback Benchmark 1280x720
http://i.imgur.com/kXbSUzQ.png

[NikosD]

I will upload in a few minutes ( I have only 1Mbps upload) three HEVC 4K 8bit clips at 100Mbps, 200Mbps and 300Mbps (it's actually the same clip encoded with the same parameters besides bitrate) from some encoding tests I have done on February 2015.

[nussman]

Nvidia results doesnt make much sense to me. Decoding and Playback FPS are almost the same? How can this be possible?
In german we say "wer misst misst Mist". ;)

What about 10bit clips on Intel Skylake with Hybrid Decoder? Has anyone made some tests yet?

[NikosD]

@Yups and anyone else who is kind enough to post their results.

No need to post the pictures because they are huge! :)

Just write the results after 5 attempts removing the first one calculate the average of the four others (approximately)

It seems like a tie in Playback mode and Skylake is around 25% faster in pure Decode mode (the real processing fixed-function power is here - without GPU processing involvement)

Nvidia results doesnt make much sense to me. Decoding and Playback FPS are almost the same? How can this be possible?
In german we say "wer misst misst Mist". ;)


It is possible because it has a lot of GPU processing power and the EVR renderer is a piece of cake for 960.

[nevcairiel]

Nvidia results doesnt make much sense to me. Decoding and Playback FPS are almost the same? How can this be possible?

Thats pretty normal if the GPU has enough power. It would only go down if the GPU isn't fast enough to process the full decoding speed for display (ie. scaling)

[NikosD]

The first huge bitrate clip is here:

4K@50fps - 100Mbps (118MB to download)
https://www.sendspace.com/file/fvgk57

Second clip:

4K@50fps - 200Mbps (238MB to download)
https://www.sendspace.com/file/usyase

Third clip: (filedropper.com "kills" links in a couple of days, so you have to download it ASAP, if you 're interested)

4K@50fps - 300Mbps (360MB to download)
http://www.filedropper.com/crowdrun2160p50300mbps

or

https://mega.nz/#!RVNy0BhB!DpsJlf5n_TJV1lptFGH4mAZblTA9-vT0P12VikS3xko

[nussman]

Mhm really? Nvidia is using fixed functions for scaling like Intel does?

[nevcairiel]

Mhm really? Nvidia is using fixed functions for scaling like Intel does?

No, but a 960 just has plenty shader power to scale 4k down, at least with the simple scaling algorithms EVR uses.

[Yups]

I don't have the 960 setup for benchmarking right now, but Yups could benchmark some of the clips I used back in January:
http://forum.doom9.org/showthread.php?p=1706902#post1706902

Specifically the high bandwidth Samsung clip would be interesting (SamsungNX1.Leaves.2160p@30fps-80Mbps - direct link (http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0235-UHD_Sample1.MP4))


Here for SamsungNX1.Leaves.2160p@30fps-80Mbps


Decode Benchmark (video size was greyed out)
http://fs2.directupload.net/images/150822/j75tgji2.png


Playback Benchmark 1280x720
http://fs2.directupload.net/images/150822/i4ri4gix.png

[NikosD]

What ?

This is almost 75% faster than 960 of Nevcairiel (!)

118 fps (960) vs 204 fps (Skylake HD 530)

[Yups]

Looks like GTX 960 suffers more than Skylake GT2 from the high bitrate. My system uses fast system memory (DDR4-3000) if this makes a difference for decoding.

[NikosD]

Looks like GTX 960 suffers more than Skylake GT2 from the high bitrate. My system uses fast system memory (DDR4-3000) if this makes a difference for decoding.

Can you test this huge H.264 bitrate 4K clip ?

http://xhmikosr.1f0.de/samples/2160p/DucksTakeOff/DucksTakeOff_2160p50.x264.CRF25.mkv

[NikosD]

Looks like GTX 960 suffers more than Skylake GT2 from the high bitrate. My system uses fast system memory (DDR4-3000) if this makes a difference for decoding.

And those two HEVC 4K 8bit clips.

The first huge bitrate clip is here:

4K@50fps - 100Mbps (118MB to download)
https://www.sendspace.com/file/fvgk57

Second clip:

4K@50fps - 200Mbps (238MB to download)
https://www.sendspace.com/file/usyase

[NikosD]

According to Yups results (with photos), the Skylake HD 530 HEVC HW decoder has this kind of performance for 4K HEVC 8 bit clips:

Decode benchmark:

BBB clip (6 Mbps)
146/167/193

LG clip (25 Mbps)
158/183/210

Samsung clip (80 Mbps)
198/204/207

Not so reasonable results, because the performance is rising with the increase of bitrate (!)

Never seen that before.

[Yups]

Can you test this huge H.264 bitrate 4K clip ?

http://xhmikosr.1f0.de/samples/2160p/DucksTakeOff/DucksTakeOff_2160p50.x264.CRF25.mkv

Decode Benchmark
Decoder Device: H264_VLD_NoFGT
Frames: 500
Time: 3,096 sec
FPS: 161,479 [150-167] fps
CPU Usage: 3 [2-4] %
GPU Usage: -

[Yups]

4K@50fps - 100Mbps (118MB to download)
https://www.sendspace.com/file/fvgk57


Decode Benchmark

Decoder Device: HEVC_VLD_Main
Frames: 500
Time: 3,027 sec
FPS: 165,160 [163-167] fps
CPU Usage: 2 [0-3] %
GPU Usage: -



4K@50fps - 200Mbps (238MB to download)
https://www.sendspace.com/file/usyase


Decode Benchmark
Decoder Device: HEVC_VLD_Main
Frames: 500
Time: 3,591 sec
FPS: 139,224 [135-140] fps
CPU Usage: 2 [2-3] %
GPU Usage: -

[NikosD]

Thank you for your results!

Did you get them by running a few times the decoding process, excluding the first run and posting the average of the rest of them ?

It seems that the H.264 engine is the same like mine, Haswell's H.264 decoder.

With my HD 4600@1.5GHz I get:

Ducks clips from XhmikosR (280Mbps)

LAV DXVA native 197/202/197

This is exactly our difference in GPU speed (1.15GHz vs 1.50GHz)

About 100Mbps & 200Mbps HEVC 8bit clips, we have to wait for the Nvidia 960 guys to post their results (if they dare of course!) :)

[nussman]

Can someone explain me please why the "GPU speed" matters here?
I thought decoding is done in co-processors?


P.S. @Yups: Could you Do some tests with HEVC 10bit sources? Skylake hybrid Decoder should be able to help for 10bit?

[NikosD]

Can someone explain me please why the "GPU speed" matters here?
I thought decoding is done in co-processors?


The coprocessor which is actually a very specific oriented ASIC and not processor in any wide form of processor, is inside Intel's iGPU and uses the exact same clock.

So, for Intel, the "GPU speed" (meaning the frequency of the clock) is the "VPU speed" the speed of the QuickSync ASIC.

[nussman]

Thanks. :)

[Yups]

Thank you for your results!

Did you get them by running a few times the decoding process, excluding the first run and posting the average of the rest of them ?

It seems that the H.264 engine is the same like mine, Haswell's H.264 decoder.


It was from the third run. Here a run with 5 starts. H.264 decode looks the same indeed. It was more than fast enough on Haswell, makes sense.

http://fs2.directupload.net/images/150822/jkld9cya.png

[P.J]

EVGA GTX 960, Win10 x64

SamsungNX1.Leaves.2160p@30fps-80Mbps (yvid_SAM_0230-ProQuality.MP4)


Decode Benchmark (video size was greyed out)
http://i.imgur.com/HHKFhdx.png


Playback Benchmark 1280x720
http://i.imgur.com/5Hx0mwZ.png

[NikosD]

EVGA GTX 960, Win10 x64

SamsungNX1.Leaves.2160p@30fps-80Mbps (yvid_SAM_0230-ProQuality.MP4)


And what about the 100 Mbps & 200 Mbps clips I uploaded ?

Can you test both with your 960 ?

[P.J]

EVGA GTX 960, Win10 x64

crowd_run_2160p50_100Mbps


Decode Benchmark (video size was greyed out)
http://i.imgur.com/5DDiv3B.png


Playback Benchmark 1280x720
http://i.imgur.com/PcJ2Va3.png

[NikosD]

As I read in Intel docs, HEVC decoder of Skylake was built for 4K60 FPS performance with 240 Mbps (!) stream.

As Yups tested, it is even faster than this.

I will upload a 300 Mbps HEVC 8bit file for Yups to test it.

[NikosD]

From P.J and Yups results, it seems that this 80 Mbps Samsung clip favors a lot Intel fixed-function HEVC decoder and is a real disaster for Nvidia fixed-function HEVC unit.

I don't know though how common scenario would that be in various HEVC encodings.

[NikosD]

@Yups and P.J

It looks like you tested different Samsung clips!

The one that Yups tested has 1394 frames and the other that P.J tested has 1016 frames !

Someone tested a wrong file!

I think Yups.

[P.J]

@Yups and P.J

It looks like you tested different Samsung clips!

The one that Yups tested has 1394 frames and the other that P.J tested has 1016 frames !

Someone tested a wrong file!

I think Yups.

Yup, I got it from here but both should be the same, 80Mbps 29.97fps:
http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0230-ProQuality.MP4

[P.J]

EVGA GTX 960, Win10 x64

crowd_run_2160p50_200Mbps


Decode Benchmark (video size was greyed out)
http://i.imgur.com/dAWaTb7.png


Playback Benchmark 1280x720
http://i.imgur.com/d0tWYJU.png

[Yups]

@Yups and P.J

It looks like you tested different Samsung clips!

The one that Yups tested has 1394 frames and the other that P.J tested has 1016 frames !

Someone tested a wrong file!

I think Yups.


I did exactly test the file nevcairiel asked and linked for.

[NikosD]

EVGA GTX 960, Win10 x64


i7-6700k HD 530 1.15 Ghz
10.18.15.4274
Windows 10 x64



@Yups and P.J

A big favor.

I want you to download and test those five 4K HEVC 8bit files and test them using DXVA Checker x64 v3.6.0, using latest LAV x64 nightly 0.65.0.47 in DXVA native mode under Win 10 x64

You have to use both modes:
Decode benchmark (no resolution) and Playback benchmark in 1280x720, five runs for each mode for each file.

PLEASE DO NOT UPLOAD SCREENSHOTS of the results!

Exclude the first run, and take the average of the other four runs for min/average/max framerate.

Your results should be like that:

Clip 1:
Decode benchmark 130/145/157
Playback benchmark 115/132/147

etc...


1) Samsung 4K@30fps-80Mbps file -0230
http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0230-ProQuality.MP4

2) Samsung 4K@30fps-80Mbps file -0235
http://imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0235-UHD_Sample1.MP4

3) Crowd_run 4K@50fps -100Mbps file
https://www.sendspace.com/file/fvgk57

4) Crowd_run 4K@50fps -200Mbps file
https://www.sendspace.com/file/usyase

5) Crowd_run 4K@50fps -300Mbps file
http://www.filedropper.com/crowdrun2160p50300mbps

Thanks!

[Aleksoid1978]

Why you guys compare HEVC 8bit decoding ?? I think 10 bit is more relevant.
On GTX 960 - HEVC DXVA 10bit decoding is faster then 8 bit. :)

[NikosD]

Because Skylake doesn't have a pure 10bit HW HEVC decoder, only hybrid.

[nussman]

Why you guys compare HEVC 8bit decoding ?? I think 10 bit is more relevant.
On GTX 960 - HEVC DXVA 10bit decoding is faster then 8 bit. :)
+1

Would like to see some tests with 10bit hybrid Decoder.

[Aleksoid1978]

And what's the point to test - if the outputs of 100 or 200 fps, what's the difference ?? :)

[NikosD]

We want to measure the scaling of performance regarding bandwidth from 80 Mbps to 300 Mbps and the difference in performance between two implementations - Skylake in HD 530 and Nvidia 960.

Isn't it obvious ?

And as a matter of fact I'm kind of tired answering to that kind of questions since 2010 when I introduced for the first time globally the concept of DXVA Benchmarking.

It's like all other types of benchmarking like CPU performance, GPU performance, GAMING performance, SW video decoder performance etc

[Aleksoid1978]

We want to measure the scaling of performance regarding bandwidth from 80 Mbps to 300 Mbps and the difference in performance between two implementations - Skylake in HD 530 and Nvidia 960.

Isn't it obvious ?

Why - what's it do ?? :)
Ok - Skylake decoding ~200 fps, GTX 960 ~100. What's next ?? :)

[NikosD]

You didn't read all of my answer...

And if you 're looking for a very practical reason, one of the first replies would be that a decoder which manages 200 FPS on a clip and another one which manages 100 fps on the same clip during benchmarking, the first one will probably consume half power decoding a 50fps clip than the latter.

A second reply would be that in transcoding environment with no renderers involved, a faster decoder could lead to a faster transcoding.

A third reply could be that a two times faster decoder - a 200 FPS decoder vs a 100 FPS decoder - means that it can support two times more simultaneous streams in a video server environment or other multi- streams decoding requirements.

Do you want a fourth reason ? ;)

[wanezhiling]

And what's the point to test - if the outputs of 100 or 200 fps, what's the difference ?? :)
The faster the decoding is, the faster the seeking is.

[NikosD]

Ahh. ..Wanezhiling just added a fourth :)

[nussman]

If Skylake Hybrid Deocder is able to play HEVC 10bit UHD streams is also quite relevant isnt it? ;)

[NikosD]

No, not for me because we already know it does and very slowly compared to pure HW decoding.

Hybrid decoding is just an intermediate "solution" between SW and pure HW decoding.

Only a 72 EUs Skylake GT4e iGPU or a 48 EUs iGPU could have some interest.

[nussman]

Hybrid Decoding is the solution for Skylake. I know it is slow, but it is important whether users can at least play "normal" (no 300mbs stuff) HEVC 10bit streams or not.

You think it is possible or not?

[NikosD]

No it's not.

It would be a huge surprise if HD 530 could play 4K60 fps 10bit in real time on anything above let's say 15 Mbps.

[nussman]

Ok. If someone with skylake is willing to test this here are some samples: http://demo-uhd3d.com/fiche.php?cat=uhd&id=46

[NikosD]

That sample is good, because it's only 23 Mbps and could confirm or not my guess.

[P.J]

This one should be better for Yups to test: http://demo-uhd3d.com/fiche.php?cat=uhd&id=97

Hybrid has terrible hiccups though.

[NikosD]

At 50 Mbps ? No, no chance at all.

Still waiting for your tests with your 960 at those 5 clips :)

[Yups]

1) Samsung 4K@30fps-80Mbps file -0230
http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0230-ProQuality.MP4

Decode= 163-168-172 [2-3%]
Playback= 132-141-145 [1-4%]



2) Samsung 4K@30fps-80Mbps file -0235
http://imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0235-UHD_Sample1.MP4

Decode= 198-205-207 [1-3]
Playback= 143-154-158 [1-3%]


3) Crowd_run 4K@50fps -100Mbps file
https://www.sendspace.com/file/fvgk57

Decode= 163-166-169 [0-2%]
Playback= 129-139-143 [2-5%]


4) Crowd_run 4K@50fps -200Mbps file
https://www.sendspace.com/file/usyase

Decode= 134-139-141 [2-3%]
Playback= 114-123-128 [2-6%]



5) Crowd_run 4K@50fps -300Mbps file
http://www.filedropper.com/crowdrun2160p50300mbps


Decode= 113-122-126 [1-5%]
Playback= 97-110-115 [3-6%]



It's a very good result I would say. SKL GT2 shader power is the first bottleneck as expected for 8 bit HEVC playback, but it's fast enough nevertheless.

[NikosD]

Perfect!

Thank you for your results.

Extremely good I would say.

It's a 4K120 FPS HEVC 8bit decoder at 300 Mbps (!)

[P.J]

EVGA GTX 960, Win10 x64:


1) Samsung 4K@30fps-80Mbps file -0230
http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0230-ProQuality.MP4

Decode= 117-118-120
Playback= 114-115-118

2) Samsung 4K@30fps-80Mbps file -0235
http://imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0235-UHD_Sample1.MP4

Decode= 119-122-126
Playback= 117-118-121

3) Crowd_run 4K@50fps -100Mbps file
https://www.sendspace.com/file/fvgk57

Decode= 129-132-134
Playback= 128-131-133

4) Crowd_run 4K@50fps -200Mbps file
https://www.sendspace.com/file/usyase

Decode= 104-107-109
Playback= 104-107-109

5) Crowd_run 4K@50fps -300Mbps file
http://www.filedropper.com/crowdrun2160p50300mbps
https://mega.nz/#!RVNy0BhB!DpsJlf5n_TJV1lptFGH4mAZblTA9-vT0P12VikS3xko

Decode= 83-93-97
Playback= 81-92-97

[P.J]

At 50 Mbps ? No, no chance at all.

Still waiting for your tests with your 960 at those 5 clips :)

That's 10bit, hope Yups gives it a try :)
http://demo-uhd3d.com/fiche.php?cat=uhd&id=97


Can you upload the 300Mbps video somewhere else like mega.nz?

[NikosD]

When I go back home I'll give it a try.

Does mega.nz allow 360 MB files ?

Is filedropper slow ?

Last time I checked it was fast enough.

[Yups]

This one should be better for Yups to test: http://demo-uhd3d.com/fiche.php?cat=uhd&id=97

Hybrid has terrible hiccups though.


Hybrid

Decode= 56-74-144 [22-36%] ~110 watts
Playback= 48-62-106 [0-28%] ~107 watts


CPU only

Decode= 83-99-156 [68-98%] ~130 watts
Playback= 62-70-81 [35-97%] ~127 watts


i7-6700k 4 Ghz @1.15V
HD 530 @1150 Mhz 10.18.15.4274


CPU only is faster in this case, but all 8 threads are running with ~90% load most of the time. I think on an ULT GT2 SKU with 4 threads only Hybrid will be faster than pure CPU.

[NikosD]

Good speed.

4K60 FPS HEVC 10bit at 50 Mbps but with the minimum values below 50fps.

Better than I thought.

What about this one ?

http://demo-uhd3d.com/fiche.php?cat=uhd&id=46

It's from Satellite.

[P.J]

Hybrid

Decode= 56-74-144 [22-36%] ~110 watts
Playback= 48-62-106 [0-28%] ~107 watts


CPU only

Decode= 83-99-156 [68-98%] ~130 watts
Playback= 62-70-81 [35-97%] ~127 watts


i7-6700k 4 Ghz @1.15V
HD 530 @1150 Mhz 10.18.15.4274


CPU only is faster in this case, but all 8 threads are running with ~90% load most of the time. I think on an ULT GT2 SKU with 4 threads only Hybrid will be faster than pure CPU.

Great :goodpost:
Would you disable HyperThreading in BIOS and try it again?


When I go back home I'll give it a try.

Does mega.nz allow 360 MB files ?

Is filedropper slow ?

Last time I checked it was fast enough.

Mega allows even big files like 10gb. Got filedropper to work.

[NikosD]

I have started uploading there too.

[nussman]

Not so bad at all. Thanks for testing.

[Yups]

Good speed.

4K60 FPS HEVC 10bit at 50 Mbps but with the minimum values below 50fps.

Better than I thought.

What about this one ?

http://demo-uhd3d.com/fiche.php?cat=uhd&id=46

It's from Satellite.


Hybrid

Decode= 64-77-117 [16-25]
Playback= 49-60-93 [13-21%]

Decoding is slightly faster in this lower bitrate video, playback roughly the same. Does a GTX 970 support Hybrid 10 bit decoding?

[NikosD]

Does a GTX 970 support Hybrid 10 bit decoding?

I am not sure.

You have to run DXVA Checker and look for HEVC_VLD_Main10 in the first page.

[nevcairiel]

Does a GTX 970 support Hybrid 10 bit decoding?

No, it does not.

[nussman]

GTX 970 is 8bit hybrid only imho.

[NikosD]

970 would definitely be faster than HD 530 in hybrid 10bit HEVC

[Yups]

It isn't support unfortunately. I will try it next time to make sure though.

[NikosD]

Mega allows even big files like 10gb. Got filedropper to work.

After fully uploading the file (it took me 1 hour) I got a message that in order to create a link to share my upload I had to create an account.

It's a free account, but I will not register to a site just to share an uploaded a file.

[P.J]

After fully uploading the file (it took me 1 hour) I got a message that in order to create a link to share my upload I had to create an account.

It's a free account, but I will not register to a site just to share an uploaded a file.

I think the 300Mbps video I have downloaded from filedropper is corrupted, 360MB.

Ok, I'll wait for new link. Mega has ability to resume upload too.

[NikosD]

No, you misunderstood.

My upload finished OK, but I have no option to share the link.

Where is the link ?

You have to make an account on that site, in order to allow you to share the upload.

You can check the integrity of the file you downloaded from filedropper by running 7 - zip or any other program capable of calculating the checksum of the file.

Run 7 - zip with CRC-32 or SHA-1 and post the number here in order to compare it with mine.

[NikosD]

CPU only

Decode= 83-99-156 [68-98%] ~130 watts
Playback= 62-70-81 [35-97%] ~127 watts


i7-6700k 4 Ghz @1.15V
HD 530 @1150 Mhz 10.18.15.4274


CPU only is faster in this case, but all 8 threads are running with ~90% load most of the time.

Your CPU results are very fast.

Using LAV x64 with the default settings of Thread to Multi-Threading = Auto I got these results.

Win 10 x64 - Core i7 4790@3.8GHz - DDR3 1600MHz - LAV x64 0.65.47

http://demo-uhd3d.com/fiche.php?cat=uhd&id=97

Decode 52/62/86 CPU avg 91% ~56W CPU package (HWiNFO)
Playback 40/49/58 CPU avg 83% ~56W CPU package (HWiNFO)

Even if I could run Haswell 4790 at 4.0GHz, our difference would be 52% faster for Skylake !

Your fast RAM could be a reason, but I think it's too much even with that.

Which is your setting for threads of LAV x64 and could you benchmark the ASTRA clip too, posting the average of CPU utilization and the setting of threads for LAV ?

Also, how do you measure the Power consumption and for what components exactly ?

It's two and a half times more than mine !!

My CPU voltage is 1.03 V.

[P.J]

No, you misunderstood.

My upload finished OK, but I have no option to share the link.

Where is the link ?

You have to make an account on that site, in order to allow you to share the upload.

You can check the integrity of the file you downloaded from filedropper by running 7 - zip or any other program capable of calculating the checksum of the file.

Run 7 - zip with CRC-32 or SHA-1 and post the number here in order to compare it with mine.

http://3.t.imgbox.com/JGn5mTT7.jpg (http://imgbox.com/JGn5mTT7)

You can create a fake email. Sendspace is good too.

[NikosD]

You are right that your download is wrong.

The CRC32 is 81DA8738

I tried first sendspace but doesn't allow 360 MB upload.

Could you try once more from filedropper ?

How did you manage to download the other two files ?

I'm open to other servers too but without registration and with easy upload method.

@Yups

Did you have problems too downloading the files ?

Is the 300 Mbps file's CRC32 the right one ?

[P.J]

Someone has uploaded it on mega for me, downloading right now:
https://mega.nz/#!RVNy0BhB!DpsJlf5n_TJV1lptFGH4mAZblTA9-vT0P12VikS3xko

Those two other videos were on sendspace not filedropper.

[NikosD]

So ?

Did you do the tests ?

The results of 300 Mbps file will be interesting, I think.

[Yups]

Your CPU results are very fast.

Using LAV x64 with the default settings of Thread to Multi-Threading = Auto I got these results.

Win 10 x64 - Core i7 4790@3.8GHz - DDR3 1600MHz - LAV x64 0.65.47

http://demo-uhd3d.com/fiche.php?cat=uhd&id=97

Decode 52/62/86 CPU avg 91% ~56W CPU package (HWiNFO)
Playback 40/49/58 CPU avg 83% ~56W CPU package (HWiNFO)

Even if I could run Haswell 4790 at 4.0GHz, our difference would be 52% faster for Skylake !

Your fast RAM could be a reason, but I think it's too much even with that.


This is a big difference indeed. Impossible to say how much the memory speed affects it, maybe I will re-test with DDR4-2133 later this week.



Which is your setting for threads of LAV x64 and could you benchmark the ASTRA clip too, posting the average of CPU utilization and the setting of threads for LAV ?

http://abload.de/img/8kxrav.png

Astra Clip:

Decode= 96-114-171 avg 72%
Playback= 66-84-87 avg 63%


Looks like CPU only handles a lower bitrate video much better. In particular CPU load is lower.


Also, how do you measure the Power consumption and for what components exactly ?


With Energy Check 3000 from the entire system.

[P.J]

So ?

Did you do the tests ?

The results of 300 Mbps file will be interesting, I think.

http://forum.doom9.org/showpost.php?p=1735592&postcount=710

[NikosD]

OK.

Here we are with all clips tested between Skylake HD 530 vs Nvidia GTX 960 using LAV x64 - DXVA native - Win 10 x64 in Decode and Playback (1280 x 720) Benchmark mode using DXVA Checker v3.60 of course (as always).


I have to thank a lot, two members of the forum, Yups and P.J for testing their hardware and giving an amount of their time for the benchmarks.

I just copied their results !

Clips:

1) LG_4K_View-the-Feeling.mp4 - 4K@30fps - 25Mbps
http://demo-uhd3d.com/fiche.php?cat=uhd&id=101

Decode:
Skylake HD 530 154/178/204
Nvidia GTX 960 145/149 /161

Playback:
Skylake HD 530 146/168/202
Nvidia GTX 960 143/148/158


2) Samsung 4K@30fps-80Mbps file -0230
http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0230-ProQuality.MP4


Decode:
Skylake HD 530 160/164/168
Nvidia GTX 960 117/118/120


Playback:
Skylake HD 530 147/155/160
Nvidia GTX 960 114/115/118



3) Samsung 4K@30fps-80Mbps file -0235
http://imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0235-UHD_Sample1.MP4


Decode:
Skylake HD 530 194/202/204
Nvidia GTX 960 119/122/126

Playback:
Skylake HD 530 174/190/201
Nvidia GTX 960 117/118/121



4) Crowd_run 4K@50fps -100Mbps file
https://www.sendspace.com/file/fvgk57


Decode:
Skylake HD 530 160/162/164
Nvidia GTX 960 129/132/134


Playback:
Skylake HD 530 143/151/164
Nvidia GTX 960 128/131/133


5) Crowd_run 4K@50fps -200Mbps file
https://www.sendspace.com/file/usyase


Decode:
Skylake HD 530 133/137/138
Nvidia GTX 960 104/107/109


Playback:
Skylake HD 530 120/129/132
Nvidia GTX 960 104/107/109


6) Crowd_run 4K@50fps -300Mbps file
http://www.filedropper.com/crowdrun2160p50300mbps
or
https://mega.nz/#!RVNy0BhB!DpsJlf5n_TJV1lptFGH4mAZblTA9-vT0P12VikS3xko


Decode:
Skylake HD 530 112/120/123
Nvidia GTX 960 83/93/97


Playback:
Skylake HD 530 102/114/119
Nvidia GTX 960 81/92/97


Comments:

1) There is a clear winner, which is Skylake in both pure Decode mode and Playback too, but with less distance for the latter.

2) Both VPUs inside GPUs, are more than capable of 4K@120fps using 100Mbps HEVC 8bit clips.

3) Samsung clips, especially the second one, favors a lot the fixed - function unit of Intel HD 530 (70% faster than Nvidia).
I don't know how common that could be, with other HEVC 8bit encodings.

4) Skylake HD 530 is close to 4K@120fps even for 300 Mbps clips. It is definitely designed for high bandwidth clips.

[NikosD]

@Yups (more) and P.J (a little)

Did you disabled ALL driver enhancements regarding video in driver's control panel ?

I always do that because especially for Intel iGPUs those enhancements add significant GPU load producing worse image most of the times and drop performance.

It's the first thing I do after installing the driver.

[P.J]

I haven't noticed any settings, which ones exactly?

Only manual settings will reduce the performance like Noise reduction/Edge enhancement...

[NikosD]

Yes, all those exactly.

They are enabled by default in Intel's drivers, I don't remember about Nvidia.

[Yups]

I didn't enable/disable anything in the driver. This is tested out of the box with Intels default settings. I will take a look later for Control Panel settings.

edit: A couple of video settings like noise reduction were enabled. Playback runs indeed faster with all settings disabled/application controlled.

[NikosD]

edit: A couple of video settings like noise reduction were enabled. Playback runs indeed faster with all settings disabled/application controlled.

I had reported that situation a few times in Intel forums and the answer was that the devs know what they are doing and they will keep enabling those features by default in their drivers.

Intel has the least powerful GPUs from all OEMs, but they keep on enabling whatever they want.

Are the results significantly faster with everything disabled or just a couple of frames ?

[NikosD]

Astonishing performance on the first DX12 game benchmark from AMD cards.

They were designed since CGN 1.0 with asynchronous compute engines (ACE), which truly shine using DX12.
On the other hand, all architecture designs of Nvidia cards before Maxwell were based on DX11.
But even the fastest Maxwell card - 980 Ti - is matched in performance by a previous generation AMD card, R9 290X!

While we are still waiting for the DX12 Win10 driver for Fermi, take a look at this:
http://arstechnica.com/gaming/2015/08/directx-12-tested-an-early-win-for-amd-and-disappointment-for-nvidia/

[P.J]

Wish Skylake had 10bit support. What exactly is Intel ClearVideo?

It's soon to comment about DX12, need more new games.

[huhn]

mantle and DX12 have a lot in common.

the main feature is the low level API.

and don't take one test to serious.

[NikosD]

What exactly is Intel ClearVideo?



Intel's ClearVideo is mainly a general term Intel used for all its video processing technologies (SW and HW).

Like PureVideo of Nvidia.

But it depends on the content of the whole concept that the term was mentioned.

Where did you see that ?



It's soon to comment about DX12, need more new games.

mantle and DX12 have a lot in common.

the main feature is the low level API.

and don't take one test to serious.

Of course is early to think about final conclusions regarding DX12 and graphics card's performance, but until we see some major redesign of Nvidia architecture - that was optimized for DX10 and DX11 - for DX12, the numerous SW teams of Nvidia can't help that much with driver's optimizations especially for pre-Maxwell cards.

You know, there is a rumour/ fact floating around that the SW developers inside Nvidia are more than HW engineers, but still they can't do much to transform the HW via SW optimizations.

[huhn]

and it is no secret that an AMD card for the same money has more raw processing power than an nvidia card.

a r9 270 and a gtx 960 have about the same gflops as raw processing power a r9 270 is way cheaper and in games the gtx 960 will clearly win.

there are other test that show that AMD gains more but not as ridiculous as this and more important they show that nvidia is gaining a lot too.

http://www.golem.de/news/vorabtest-directx-12-macht-was-es-soll-1502-112228.html

i would love to see AMD gain some market share but no they didn't add 10 bit HEVC decoding and they didn't add HDMI 2.0 in there new GPUs...

[NikosD]

i would love to see AMD gain some market share but no they didn't add 10 bit HEVC decoding and they didn't add HDMI 2.0 in there new GPUs...

And Intel did the exact same thing like AMD and the new Skylakes.

So, they probably both thinking that 4K60 and BluRay UHD are not very popular yet or even existent.

[huhn]

but people need HDMI 2.0 right now...

[P.J]

Intel's ClearVideo is mainly a general term Intel used for all its video processing technologies (SW and HW).

Like PureVideo of Nvidia.

But it depends on the content of the whole concept that the term was mentioned.

Where did you see that ?

Does ClearVideo support pixel adaptive deinterlacing?
What about Celeron/Pentium series that don't have ClearVideo?

[huhn]

there are some celeron and pentium without clearvideo that didn't have hardware decoder.

deinterlacing should be supported by all but not 100 % sure.

[NikosD]

ALL Celeron/Pentium processors have fixed-function HW decoders and Haswell's Pentium/ Celeron have full QuickSync support, meaning the encode path too.

And they all have adaptive deinterlacing of course.

[huhn]

sandy had the problem but haswell and newer shouldn't be affected anymore.

http://semiaccurate.com/2011/03/21/intels-sandy-bridge-celerons-video-features-fused-off/

Some of the missing features such as InTRU 3D and support for WiDi is sort of understandable on a budget CPU like this, but not only has Intel dropped support for Clear Video HD, but even standard Clear Video isn’t supported which means that the graphics core can’t offload the CPU when it comes to video playback, features you’d expect to be part of an entry level CPU as this is where processor offloading makes the most sense.

[NikosD]

It's dead wrong.

ALL Pentium/Celeron including Sandy have the fixed-function decoder enabled.

[nevcairiel]

sandy had the problem but haswell and newer shouldn't be affected anymore.

http://semiaccurate.com/2011/03/21/intels-sandy-bridge-celerons-video-features-fused-off/

You should read the update to the story, Intel only removed QuickSync on those, decoding is supported. Some of the fixed-function video enhancements have also been disabled, but noone wants those anyway!

[huhn]

good to know.

this was actually an non existing issue people talked about in the past.

so that leaves only the first ATOM with no hardware decoder.

[Yups]

Retested with all image enhancements disabled in the driver.



1) Samsung 4K@30fps-80Mbps file -0230
http://www.imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0230-ProQuality.MP4

Decode= 160-164-168 [2-3%]
Playback= 147-155-160 [3%]



2) Samsung 4K@30fps-80Mbps file -0235
http://imaging-resource.com/PRODS/samsung-nx1/FULLRES/yvid_SAM_0235-UHD_Sample1.MP4

Decode= 194-202-204 [2%]
Playback= 174-190-201 [2%]


3) Crowd_run 4K@50fps -100Mbps file
https://www.sendspace.com/file/fvgk57

Decode= 160-162-164 [2%]
Playback= 143-151-164 [3%]


4) Crowd_run 4K@50fps -200Mbps file
https://www.sendspace.com/file/usyase

Decode= 133-137-138 [2%]
Playback= 120-129-132 [4%]



5) Crowd_run 4K@50fps -300Mbps file
http://www.filedropper.com/crowdrun2160p50300mbps


Decode= 112-120-123 [3%]
Playback= 102-114-119 [4%]



Biggest difference for video 2. Playback closed in. Decode is slightly slower than my previous test. I did install Bitdefender, maybe it slows down the test a bit.

[NikosD]

Thanks!

I'll update the overall post above.

I'm pretty sure that 10bit hybrid results will be better too.

I'm very interested in your RAM downgrade results to 2133 MHz (default for Skylake) to see how lower could be in SW decoding and maybe HW too compared to 3000 MHz.

[P.J]

Thanks!

I'll update the overall post above.

I'm pretty sure that 10bit hybrid results will be better too.

I'm very interested in your RAM downgrade results to 2133 MHz (default for Skylake) to see how lower could be in SW decoding and maybe HW too compared to 3000 MHz.

Yes, those would be great to test + No Hyperthreading :)
I wonder if i5 6600K can handle 10bit well too.

[NikosD]

Retested with all image enhancements disabled in the driver.


Can you retest this too, in order to complete the picture:

1) LG_4K_View-the-Feeling.mp4 - 4K@30fps - 25Mbps
http://demo-uhd3d.com/fiche.php?cat=uhd&id=101

[Yups]

LG_4K_View-the-Feeling.mp4

decode= 154-178-204 3%
playback= 146-168-202 4%

[Yups]

LG_4K_View-the-Feeling.mp4

decode= 154-178-204 3%
playback= 146-168-202 4%


After a bios update:


decode= 159-184-212 3%
playback= 148-170-203 3-4%

[NikosD]

The platform is very young and those fluctuations are expected.

I'll keep the previous results of clip 1 in order to be inline with the others.

Unless you want to rerun the tests for the rest too or you tell me that the others didn't change.

[NikosD]

i7-6700k HD 530 1.15 Ghz
10.18.15.4274
Windows 10 x64


@Yups

Using latest DXVA Checker v3.70 and your HD 530, if you go to "Check Decoders" -> "Select Format" and you choose VP8 or VP9, are there any other decoders besides LAV ?

I'm interested in Intel MFT VP8/VP9 decoders that using previous DXVA Checker v3.60 were enumerated by choosing VP8 and VP9 formats.

Also, if you have a VP8 or even VP9 sample (any sample) can you check with your system and latest DXVA Checker v3.70, if any Intel MFT decoders appear when you choose "Select Media File..." and test a VP8/VP9 file ?

Thanks!

[P.J]

Are upcoming 4K Bluray discs 10bit or 8bit?

[sneaker_ger]

10 bit (UltraHD TV and UltraHD streaming are also often 10 bit already)

[Yups]

@Yups

Using latest DXVA Checker v3.70 and your HD 530, if you go to "Check Decoders" -> "Select Format" and you choose VP8 or VP9, are there any other decoders besides LAV ?

I'm interested in Intel MFT VP8/VP9 decoders that using previous DXVA Checker v3.60 were enumerated by choosing VP8 and VP9 formats.

Also, if you have a VP8 or even VP9 sample (any sample) can you check with your system and latest DXVA Checker v3.70, if any Intel MFT decoders appear when you choose "Select Media File..." and test a VP8/VP9 file ?

Thanks!


No decoder listed. I don't have VP8/9 samples.


http://abload.de/img/unbenanntjbj9p.png

[NikosD]

But the image shows clearly Intel's VP9 decoder :)

If you select VP8 format, does Intel's VP8 MFT decoder appears ?

I'll upload two samples later, one VP8 and one VP9 to test the "Select Media File"

[Yups]

VP8 shows exactly the same.

[NikosD]

Weird...It should say VP8 according to the previous version of DXVA Checker and according to the installed decoders.
Unless of course it is somehow common.

[P.J]

Using latest DXVA Checker v3.70 and your HD 530, if you go to "Check Decoders" -> "Select Format" and you choose VP8 or VP9, are there any other decoders besides LAV ?

I did it for HEVC but got no HEVC_VLD_Main10 for LAV:

http://i.imgur.com/fEslJSK.png

[sneaker_ger]

Does it only show native DXVA? Then you might need a recent nightly version of LAV.
http://files.1f0.de/lavf/nightly/

[P.J]

Does it only show native DXVA? Then you might need a recent nightly version of LAV.
http://files.1f0.de/lavf/nightly/

Still no luck.

[NikosD]

No decoder listed. I don't have VP8/9 samples.




Here it is a VP8 sample:
https://www.sendspace.com/file/be21aw

And a VP9 from Youtube:
https://www.sendspace.com/file/v95vjb

[oscarbg]

After a bios update:


decode= 159-184-212 3%
playback= 148-170-203 3-4%

can you please post full system specs (motherboard model, bios versions,etc..)

[oscarbg]

Guys isn't time to test HW accelerated HEVC encode?
It is present on Skylake and NV Maxwell Gen 2 GPUs (yes even my GTX 970 which doesn't support HEVC decode)..
One way to test will be ffmpeg which supports using NVENC for Nvidia support and QuickSync for Intel..
We should make standard binary and command line args with a standard video input too similar to h264 benchmark (Tech ARP - x264 HD Benchmark Ver. 5.0.1)..
Then let the benchmarks begin..

[NikosD]

I have been thinking of this concept since last August when I purchased my Core i7 4790.

But most people are interested in SW encoding and particularly SW quality encoding, not speed.

So, a very specific benchmark test regarding speed of HW encoding and expecially HEVC, which is so immature right now, it doesn't make a lot of sense to me.

But you are free to open such a thread, I don't think there is something like that in this forum.

I know only a quality comparison thread between x264 and x265.

[P.J]

It's kinda useless since SW encoding provides better quality and nobody wants to encode in real-time on PC.

[huhn]

It's kinda useless since SW encoding provides better quality and nobody wants to encode in real-time on PC.

real time encoding is very important for live streaming such as streaming computer games to TVs and things like that.

if HEVC hardware encoding is better than the old AVC hardware encoder they would be really useful for such things.

the pictures i saw from the nvenc HEVC encoder wasn't good at all.

[NikosD]

I did it for HEVC but got no HEVC_VLD_Main10 for LAV:



Talking with DXVA Checker's developer, he told me that the new 3.7.2 version has fixed that (not released yet)

No decoder listed. I don't have VP8/9 samples.


The VP9and8 DXVA device decoder has been removed because it's not supported by Intel (yet or ever!)

Also the developer of DXVA Checker told me, that it's impossible to use DXVA Checker for VP9/VP8 samples with Intel MFT decoder, because it simply is not working (yet or ever!)

[NikosD]

For anyone interested in testing, I encoded a 8K (7680x4320) HEVC clip at 60fps in both 10bit and 8bit versions.

Of course, you can't HW accelerate the decoding of those clips even if you have a 960 card, due to high resolution.

You need a very fast CPU.

The clips are just 1 sec and very small.

Clip 1- HEVC 4320p60fps -10bit


Clip 2 - HEVC 4320p60fps -8bit





@Yups

Back on February of 2015, a few months ago, I have posted two 8K (!) HEVC clips and a lot of people could think that I'm crazy to post such high-resolution HEVC clips, since not even 4K is largely populated yet.

BUT as I read in Intel forums, Skylake supports in HW the 8K resolution for HEVC (!!) (not for H264)

I uploaded again those two clips in 10 bit (that you can't test in pure HW) and a 8bit HEVC 8K clip that I would like you to test.


The clips are just 1 sec and very small.


Clip 1- HEVC 4320p60fps -10bit
https://www.sendspace.com/file/wglrw4

Clip 2 - HEVC 4320p60fps -8bit
https://www.sendspace.com/file/difznd


I would like to see some benchmarks in both modes (Decode and Playback) with DXVA native using DXVA Checker.

Also, it would be interesting to test Skylake CPU and hybrid GPU with 8K HEVC 10bit clip.

Thanks!

[Yups]

This one should be better for Yups to test: http://demo-uhd3d.com/fiche.php?cat=uhd&id=97

Hybrid has terrible hiccups though.

Hybrid

Decode= 56-74-144 [22-36%] ~110 watts
Playback= 48-62-106 [0-28%] ~107 watts


CPU only

Decode= 83-99-156 [68-98%] ~130 watts
Playback= 62-70-81 [35-97%] ~127 watts


i7-6700k 4 Ghz @1.15V
HD 530 @1150 Mhz 10.18.15.4274


CPU only is faster in this case, but all 8 threads are running with ~90% load most of the time. I think on an ULT GT2 SKU with 4 threads only Hybrid will be faster than pure CPU.



Because someone asked, here 4 CPU cores without HT.


Hybrid

Decode= 56-76-144 [37-78%, avg 51%] ~102 watts
Playback= 51-62-108 [28-54%, avg 41%] ~97 watts


CPU only

Decode= 63-72-99 [72-99%, avg 92%] ~105 watts
Playback= 58-70-98 [73-98%, avg 90%] ~105 watts

[NikosD]

Did you see my post above with 8K HEVC (10bit/8bit) clips ?

[Yups]

@Yups

Back on February of 2015, a few months ago, I have posted two 8K (!) HEVC clips and a lot of people could think that I'm crazy to post such high-resolution HEVC clips, since not even 4K is largely populated yet.

BUT as I read in Intel forums, Skylake supports in HW the 8K resolution for HEVC (!!) (not for H264)

I uploaded again those two clips in 10 bit (that you can't test in pure HW) and a 8bit HEVC 8K clip that I would like you to test.


The clips are just 1 sec and very small.


Clip 1- HEVC 4320p60fps -10bit
https://www.sendspace.com/file/wglrw4

Clip 2 - HEVC 4320p60fps -8bit
https://www.sendspace.com/file/difznd


I would like to see some benchmarks in both modes (Decode and Playback) with DXVA native using DXVA Checker.

Also, it would be interesting to test Skylake CPU and hybrid GPU with 8K HEVC 10bit clip.

Thanks!


Not sure how reliable a 1 second video is regarding fps but the 8 bit HEVC sample is working in hardware. 60 fps decode, 55 fps playback with low CPU load. Pure CPU decoding/playback is much slower. The 10 bit sample crashes in DXVA Checker.

[NikosD]

Thanks!

You confirm that is working and that's very good.

Does the 10bit video crash with MPC-HC or other player or DXVA checker only ?

Does it crash also with MS MFT using decode mode (the only usable HW mode) or Cyberlink H265 DXVA/OCL decoder ?

I want to confirm that it's a DXVA Checker problem and not decoders problem.

[NikosD]

Next DXVA Checker will check for 8K or 8KFUHD as it is called (Full Ultra HD)

[Yups]

Crash/hangs happens while loading/searching lav video decoder. There is no crash when I disable hardware decoder in lav settings.

[NikosD]

Crash/hangs happens while loading/searching lav video decoder. There is no crash when I disable hardware decoder in lav settings.

Can you try MS MFT H265 decoder (with "Enable DXVA" setting) in "Decode Benchmark" mode for 10bit 8K HEVC ?

Thanks!

[P.J]

Because someone asked, here 4 CPU cores without HT.


Hybrid

Decode= 56-76-144 [37-78%, avg 51%] ~102 watts
Playback= 51-62-108 [28-54%, avg 41%] ~97 watts


CPU only

Decode= 63-72-99 [72-99%, avg 92%] ~105 watts
Playback= 58-70-98 [73-98%, avg 90%] ~105 watts

:thanks:

[foxyshadis]

Completely off-topic posts about DX12 gaming moved to the PC hardware forum. At least keep this to decoding in some way, even if it's not HEVC specific anymore.

[P.J]

http://www.thedailynote.net/2014/05/why-4k-video-looks-better-on-1080p-monitors.html

[NikosD]

IIRC DXVA accelerates only 4:2:0 clips

[leonccyiu]

@Yups

Back on February of 2015, a few months ago, I have posted two 8K (!) HEVC clips and a lot of people could think that I'm crazy to post such high-resolution HEVC clips, since not even 4K is largely populated yet.

BUT as I read in Intel forums, Skylake supports in HW the 8K resolution for HEVC (!!) (not for H264)

I uploaded again those two clips in 10 bit (that you can't test in pure HW) and a 8bit HEVC 8K clip that I would like you to test.


The clips are just 1 sec and very small.


Clip 1- HEVC 4320p60fps -10bit
https://www.sendspace.com/file/wglrw4

Clip 2 - HEVC 4320p60fps -8bit
https://www.sendspace.com/file/difznd


I would like to see some benchmarks in both modes (Decode and Playback) with DXVA native using DXVA Checker.

Also, it would be interesting to test Skylake CPU and hybrid GPU with 8K HEVC 10bit clip.

Thanks!

Wow! Does the Skylake Core M with Intel HD Graphics 515 Also do 8k HEVC?

http://www.notebookcheck.net/Intel-HD-Graphics-515.149941.0.html

But since there's no 10 bit HEVC and 8k will likely have frame rates of at least 60, not to mention other codecs like vp9 which a core m isn't going to decode 4k60 in, not sure how useful it is. I heard maybe Kaby Lake and Gen10 will have 10bit HEVC decode.

[leonccyiu]

Not sure how reliable a 1 second video is regarding fps but the 8 bit HEVC sample is working in hardware. 60 fps decode, 55 fps playback with low CPU load. Pure CPU decoding/playback is much slower. The 10 bit sample crashes in DXVA Checker.

If it's 55fps playback for the 60fps videos, would that be frames being dropped? and therefore 8k60 is not possible unlike skylake?

[Yups]

Probably Kaby Lake includes 10bit HEVC but because it's in 14 nm process again, maybe they have to cut something else to fit 10 bit HEVC.


SKL 4+2 is only ~120 mm² big...plenty of die size headroom for improvements on 14nm.

[P.J]

Because someone asked, here 4 CPU cores without HT.


Hybrid

Decode= 56-76-144 [37-78%, avg 51%] ~102 watts
Playback= 51-62-108 [28-54%, avg 41%] ~97 watts


CPU only

Decode= 63-72-99 [72-99%, avg 92%] ~105 watts
Playback= 58-70-98 [73-98%, avg 90%] ~105 watts

Would you test with DDR4 2133 too? :o

[P.J]

http://www.computerbase.de/2015-09/amd-radeon-r9-nano-test/6/

[NikosD]

Tremendous processing power of GPU shaders could make you think that R9 Nano has pure HEVC HW decoder, but it doesn't.

Or more carefully I would say it shouldn't.

It's two times faster of the Nvidia 980 in 4K HEVC decoding.

[nevcairiel]

But Nano and Fury (X) do have a HEVC hardware decoder - limited to 8-bit only, but still hardware.
I dare say a hybrid decoder couldn't reach such speed. It would be CPU limited before the shaders could even all be used.

[Sulik]

I dare say a hybrid decoder couldn't reach such speed. It would be CPU limited before the shaders could even all be used.

Yeah, all hybrid decoders are fundamentally limited by CABAC on the CPU: they really only work well at low bitrates. If you go to low enough bitrates, the hybrid decoders can probably surpass pure HW decoders, but that doesn't mean much.
That's similar to software vendors showcasing their decoders -> anyone remembers the DivX 4K streams at 2Mbps ? :)

[NikosD]

I remembered commenting here the presence of HEVC decoding block inside Fury cards and the lack of HDMI 2.0 but probably because I have never seen benchmark results of HEVC for AMD I was under the idea that only Carizzo can decode in HW HEVC format.

Isn't there someone with a Fury card or a Carrizo laptop to do some HEVC tests, BTW ?

For the hybrid results I agree 100% with Sulik.

All of my tests show a full usage of shaders for low powered GPUs like HD 4600 and low bitrate clips.

For those low bitrate HEVC clips a very powerful card like R9 Nano - if it didn't have a HEVC fixed function block - could reach or even surpass a fixed function unit which can clearly show its teeth in high and very high bitrates.

[NikosD]

A new graphics department from AMD, called Radeon Technologies Group could easily be sold. That or AMD itself.

And who is gonna buy them, AMD and Radeon Group ?
Google ? Apple ? Samsung ? Or even Qualcomm ?

All I know is that I'm gonna buy after six years (!) a discrete card and it's going to be AMD or Radeon Technologies Group (or whatever they called)

I'm gonna buy AMD Greenland next-gen card on 2016, especially if it has HEVC 10bit and hopefully VP9 in HW.

http://anandtech.com/show/9612/amd-reorganizes-graphics-division-radeon-whole-once-more-led-by-raja-koduri

[baii]

Any chance to update SW and hybrid result for high bit rate clips? or they just struggle?
CPU hw decoder seem so gimped as they don't do 10bit where 960 is too weak to do needi3 ~

[NikosD]

I think it"s pointless.

CPU and hybrid decoders are not built for such tasks, like high bitrate decoding, although the iGPU of Skylake did better than I thought so on 10bit HEVC.

[aufkrawall]

Intel seems to know what they are doing, even HEVC 8 bit hybrid decoder of Haswell wasn't slow at all.
They probably get a lot of useful findings from their Quick Sync development.
Since they basically have yearly hardware updates, there's not much pressure to put everything in hardware at once.

[huhn]

so you have to buy every year?

[foxyshadis]

You buy as soon as something comes out that has improved performance enough to interest you. The dilemma that's faced hardware upgraders forever.

[huhn]

for the next years all we need is a hardware decoder that can decode 10 bit UHD 60 HZ

but adding 8 bit HEVC is like adding nothing at all.
that's why i'm not happing at all what intel is thinking.
same for AMD no HDMI 2.0 no 10 bit HEVC.

it's just terrible that you have to buy an nvidia 960 or 950.

and of cause nothing is older than the hard you just brought.

[aufkrawall]

Well, don't want to sound like a smart ass, but that's just the result of market economy with weak competition (AMD CPUs/APUs).
They are giving customers a reason to buy also next year when they probably release Kabylake (?) way before AMD Zen.

[baii]

Maybe a silly question, do any of the hw decoder run in 32bit player?

Sent from my SM-T700 using Tapatalk

[NikosD]

ALL of them.

[Yups]

Well, don't want to sound like a smart ass, but that's just the result of market economy with weak competition (AMD CPUs/APUs).
They are giving customers a reason to buy also next year when they probably release Kabylake (?) way before AMD Zen.


It's not even clear if Kabylake brings 10 bit support and HDMI 2.0. If this is just a Skylake refresh like Devil's Canyon, means Skylake CPU+Gen9 GPU, it is unlikely.

[aufkrawall]

Well, then Intel would probably be outperformed by next year's AMD / Nvidia GPU generation.
Why would they allow this to happen? I can hardly believe that they are failing technically to implement native HEVC 10 bit decoding when Haswell was so far ahead, in comparison to NV and especially AMD.

[Yups]

In regards to technically, dedicated GPUs will always have a big time advantage over a CPU with GPU, an entire platform with new chipsets/sockets and all the neccessary drivers because of a much longer development process with much longer evaluating times. There is a reason why AMDs GPU in their APUs was a generation behind their dedicated GPUs in the past. GPU only is much more flexible.

[NikosD]

DXVA checker v3.80 is out with 8K test and LAV Video 10bit HEVC support in "Select Format" option

[aufkrawall]

In regards to technically, dedicated GPUs will always have a big time advantage over a CPU with GPU, an entire platform with new chipsets/sockets and all the neccessary drivers because of a much longer development process with much longer evaluating times. There is a reason why AMDs GPU in their APUs was a generation behind their dedicated GPUs in the past. GPU only is much more flexible.
They will have had three years time to upgrade their Haswell VPU to native HEVC 10 bit decoding.
Which should be quite enough, since NV has it since January 2015.

AMD is terribly short on r&d financial resources, I wouldn't compare this to Intel.

[foxyshadis]

I know most people use 3rd party players, but saw this tidbit in a leaked Microsoft Win10 release notes for last week and figured someone might be interested:

Media: We made changes to ensure that a high efficiency video coding (HEVC) can be played in the Movies & TV app when decoder bit depth downsampling is recommended by a Direct3D version 11 user-mode driver.

In other words, 10-bit file with 8-bit rendering was broken, now it's not.

[wanezhiling]

I bought a Surface 3 (not Pro), x7 z8700 is capable of 8-bit hevc, indeed.

http://i3.tietuku.com/c4aaf96029a52ec8.jpg

http://i3.tietuku.com/7d83b4807bd6a620.jpg

http://i3.tietuku.com/6b66755d815712d5.jpg

[P.J]

I bought a Surface 3 (not Pro), x7 z8700 is capable of 8-bit hevc, indeed.

http://i3.tietuku.com/c4aaf96029a52ec8.jpg

http://i3.tietuku.com/7d83b4807bd6a620.jpg

http://i3.tietuku.com/6b66755d815712d5.jpg

https://www.google.com/search?es_sm=93&q=print+screen+key&spell=1&sa=X&ved=0CBoQvwUoAGoVChMIp8Cnn_SDyAIVKZ3bCh1cOw66

Try to benchmark some of the videos here and post the result.

[wanezhiling]

Try to benchmark some of the videos here and post the result.
Those clips are too slow for me to download... you know there is a 'firewall' in my country..:o

Well I tested one easy 8-bit hevc clip and it was ~6% slower than hd 530, maybe atom is just too weak in some other chains.

[leonccyiu]

I bought a Surface 3 (not Pro), x7 z8700 is capable of 8-bit hevc, indeed.

http://i3.tietuku.com/c4aaf96029a52ec8.jpg

http://i3.tietuku.com/7d83b4807bd6a620.jpg

http://i3.tietuku.com/6b66755d815712d5.jpg

Oh wow! Even an Atom can decode HEVC 8 bit in 8k, and do 10 bit files work now after being downsampled to 8 bit?

It'd be nice if hdmi 2 was more prevalent, it's been years since it's been available, even displayport would be nice.

I am going a bit off topic here, but since VP9 is google's codec for Youtube which goes up to 4k60.
I've seen on anandtech that the atom has hardware vp9 decoding.
http://www.anandtech.com/show/9219/the-surface-3-review/4
Although I think it's probably just hybrid like on broadwell and skylake, on your dxva checker I don't see vp9.

I don't think a vp9 dxva protocol exists, I am using a pentium g3258 on a z87 mb so broadwell or skylake would be expensive upgrades, whereas an i5-4690k can do 4k60vp9 in software but comes close to maxing the cpu.

Is there any browser such as chrome that can use the hybrid vp9 decode on Intel Graphics?

[NikosD]

Oh wow! Even an Atom can decode HEVC 8 bit in 8k, and do 10 bit files work now after being downsampled to 8 bit?

http://www.anandtech.com/show/9219/the-surface-3-review/4


Back on May of 2015, I had commented that article asking about HEVC.

I copy the comments here:


NikosD - Monday, May 04, 2015 - link
That was a very good chance to test HEVC decoding capability of Cherry Trail.

How fast is it ?
Is it fast enough to decode 4K@60 fps ?

Does it support 10 bit HEVC decoding in hardware ?
REPLY
Brett Howse - Monday, May 04, 2015 - link
This is something I wanted to try out before the review but I ran out of time. I'll get to this sometime this week and give you an update.
REPLY
Ryan Smith - Monday, May 04, 2015 - link
Intel's specs specifically list Cherry Trail at topping out at 4Kp30 (i.e. Level 5).
REPLY
NikosD - Tuesday, May 05, 2015 - link
Thank you both.
REPLY
Brett Howse - Wednesday, May 06, 2015 - link
I tried a 4K60p 8 bit, and it could almost play it back, but 4k30p 10 bit was a no go. But I'm no expert on HEVC that's Ganesh's field.


So it can do (almost) a 4K60p 8 bit.

Atom has 16 EUs and in playback mode that could be an issue for scaling of high resolution clips such as 4K or even worse 8K.

[sneaker_ger]

and do 10 bit files work now after being downsampled to 8 bit?
What is that even supposed to mean?

[NikosD]

Well I tested one easy 8-bit hevc clip and it was ~6% slower than hd 530, maybe atom is just too weak in some other chains.

If you manage to download even one clip, it would be interesting to show your results in both "Decode benchmark" mode and "playback benchmark" mode along with your GPU clock in order to compare with HD 530 using LAV native and DXVA Checker of course.

P.S
Don't forget to disable EVERYTHING in driver's control panel of Windows, because especially for weak GPUs like those iGPUs from Intel, play a significant role in "playback mode" performance.

[P.J]

and do 10 bit files work now after being downsampled to 8 bit?

:confused:

[daWsOn_s]

Do you know if Broadwell-U CPUs can playback 4k hevc with hybrid decoding?

[NikosD]

Yes it does.

Broadwell supports hybrid decoding of 4K HEVC 8bit/10bit and VP9.

[daWsOn_s]

Yes it does.

Broadwell supports hybrid decoding of 4K HEVC 8bit/10bit and VP9.

I meant, is playback smooth? Meaning is there a limit for the bitrate when the hybird decoding won't be enough? I'll be using 8bit 4k 80Mbit.

[NikosD]

I don't have experience on my own and I haven't seen a lot of discussions around this processor.

Broadwell for desktop is almost non existant, but Broadwell-U is an older processor used in laptops and NUC products.

I read that there are several GPU configurations:

28W with GT3, Iris 6100 Graphics (48 Execution Units)
15W with GT3, HD 6000 Graphics (48 Execution Units)
15W with GT2, HD 5500 Graphics (23 Execution Units for low i3, 24 for others)
15W with GT1, HD (Broadwell) Graphics (12 Execution Units)

80 Mbps is a huge bandwidth for hybrid decoding especially for HEVC 4K, but if you aim for 48EUs iGPU model with HD 6000/6100 you could have a chance.

I don't think there is a hardcoded limit for bitrate, but if your GPU can't decode the clip or the CPU can't keep up feeding the GPU by decoding its part, then there is nothing that can help you.

It would be better to find some answers from someone who has actually tried it though.

[daWsOn_s]

I see, I wanted a new laptop with hevc support but skylake machines are much more expensive so I wanted to save going with broadwell...seems like I could get a crippled support...

[leonccyiu]

I meant, is playback smooth? Meaning is there a limit for the bitrate when the hybird decoding won't be enough? I'll be using 8bit 4k 80Mbit.

On Notebookcheck reviews of the Retina 13 and Macbook with Broadwell they briefly report on h.265 decode abilities.

http://www.notebookcheck.net/Apple-MacBook-Pro-Retina-13-Early-2015-Notebook-Review.139621.0.html

For the Retina 13 which uses an i5U with Iris 6100 graphics it plays back smoothly. There is a graph which shows ~37% cpu usage. I think the Iris 6100 is 50% faster than the hd5500.

For the Core M macbook they say a 4k h265 clip at 2.7mbps is a challenge.

[leonccyiu]

https://blogs.gnome.org/rbultje/2015/09/28/vp9-encodingdecoding-performance-vs-hevch-264/

On this page where several decoders are compared for the same quality video, I notice a faster decoder than ffh265 called openhevc. I can't seem to find a player for windows on the net. I believe lav hevc decoder is derived from ffh265? I've probably got this wrong somewhere

[CruNcher]

@NikosD

Can Cyberlinks Decoder still be used i dont get it to connect :( ?

Though PowerDVD 15 itself shows catastrophical Performance results

here is one of my own results currently with VP6 on the GTX 970 though not DXVA and more Realtime Processing (Multithreading Stability Overhead) oriented then a more synthetic benchmark, i try to task every part of the system in a real Workflow scenario (gonna add the Realtime Encoder CUDA, and ASIC next and task them) :)

and slowly highering the decoding complexity on the DSP Engine and adding more Tasks for the OS to handle @ the same time to bring each slowly to it's knees and measure which dies faster ;)


http://i1.sendpic.org/t/zp/zpUqUHt2JSFa8Zmuy6W5HtpFRzU.jpg (http://sendpic.org/view/1/i/hWRpALVaHzBYpeKDjz8pMj5oqtU.png)
http://i2.sendpic.org/t/hb/hbOOATZ7SduB3qmF6324G4Klpto.jpg (http://sendpic.org/view/2/i/pK45vLwHzJL6Qy7PC15hNYN38jZ.png)
http://i1.sendpic.org/t/lR/lR0rzwR2sCVJQAJHhneUnWPDOGF.jpg (http://sendpic.org/view/1/i/uyfN9USX5QYFdmZYVnf72GB6fpQ.png)

It also shows how nicely Firefox Engine has Improved running in the background while testing and doesn't causing noticeable interference's anymore :)




8 Bit results via CUDA (DXVA2 mix)

Lets go insane (it most probably overloads especially the USB 3 Drive, downloading on the same drive @ the same time) ;)

http://i2.sendpic.org/t/vm/vmBxrHHioMGKao6Wdikqxyhi1Uv.jpg (http://sendpic.org/view/2/i/8diVdHsL1rSRGM5wBWddR5qYNDb.png)

Lets go sane :) 2 sane ones even on DX9 renderer are doable

http://i1.sendpic.org/t/3S/3SoCiPOj79WUuMUFLNbnQZliHRU.jpg (http://sendpic.org/view/1/i/goyQPc39YEeSoEEbk5fuEwhf6gO.png)

Streams come from the same USB 3 Hard Drive Realtime Preview test.

http://i2.sendpic.org/t/dC/dC9RpVliGeLEbxaLVuuAE7NlF1T.jpg (http://sendpic.org/view/2/i/65AOb6l1FPkU6zZODWrJYoxJqFD.png)

System = Sandy Bridge I5-2400 on Windows 7 Nvidia WDDM 1.1



Totally unstable playback results PowerDVD 15 the whole thing is still 32 Bit :(

http://i2.sendpic.org/t/yb/ybLD6xcBGS9PhYKagjE0K5dWeC7.jpg (http://sendpic.org/view/2/i/vFNfGe0pYKq0YtYWp3C8YbiDajr.png)

[NikosD]

I see, I wanted a new laptop with hevc support but skylake machines are much more expensive so I wanted to save going with broadwell...seems like I could get a crippled support...

It looks like that the potential buyer of a Core M (Broadwell - Y) should be extra careful of the exact product.

According to anadtech.com, the Intel Core M 5Y10c with iGPU HD 5300 doesn't support hybrid HEVC at all! (not even 8bit)

http://www.anandtech.com/show/8303/zbox-sphere-oi520-plus-zotacs-take-on-the-nuc/5

Of course this kind of CPU, would never find a place in a laptop.

[nussman]

Broadwell and skylake are pretty much useless for HEVC, because no 10bit support.
I would wait or buy something cheap.

1080p 8bit Clips with "normal" bitrates can be decoded by a modern CPU.

[CruNcher]

Seeing Intel lose here i can hardly believe it

It's expected that AMD would take the lead with its more capable graphics core and UVD engine, but by what margin?

Not really

https://youtu.be/lliDV585AT0?t=126

But i guess it's a business decision saving the Space for more important transistors for the current timeframe of content distribution comparable to Nvidias sometimes strange looking decisions on the VPx layer.

They practically even showed that @ IDF they see no need for it in 2015 so why waste space now, it makes perfect sense that AMD though can still compete with UVD and still maintain better 3D Performance says a lot but energy efficiency wise it says not much.

Most Reviewers though never correctly look into these things most of the times ;)

Next year we will see all of them supporting 10 bit like the ARM SOCs do already this year.

Amd is really catching up on Intels Shrinking run if Intel doesn't slowly improves the GPU efficiency i guess it will be a full win for AMD next year maybe even a full Win in 2 areas vs Nvidia and Intel.

https://www.youtube.com/watch?v=9C9x-TB_0Ro

i guess we can believe these PR though @ which TDP ?

https://www.youtube.com/watch?v=Rp1nzQVAZKE

HSA (Fusion) seems to slowly pay off

https://www.youtube.com/watch?v=K6gqBE2_RxM

On the other side see how much load AMD puts on Intels Quicksync 64 Mbps that's far off from a Internet Streaming scenario that would make sense this year even in 8 bit ;)

Pretty well optimized

https://www.youtube.com/watch?v=26kF8vJO3gc

[Yups]

It looks like that the potential buyer of a Core M (Broadwell - Y) should be extra careful of the exact product.

According to anadtech.com, the Intel Core M 5Y10c with iGPU HD 5300 doesn't support hybrid HEVC at all! (not even 8bit)

http://www.anandtech.com/show/8303/zbox-sphere-oi520-plus-zotacs-take-on-the-nuc/5

Of course this kind of CPU, would never find a place in a laptop.


August 1, 2014



Outdated...

[mariush]

...


You can read a lot of technical information about Carrizo and its specs in the Anandtech article: http://www.anandtech.com/show/9319/amd-launches-carrizo-the-laptop-leap-of-efficiency-and-architecture-updates

15w TDP, up to 8.3 hours for h264 playback or 300 minutes for hevc 4k 60fps (with a 50wh battery), lots of power efficiency tweaks...

Page 6 has a lot of details about UVD6 :

http://www.anandtech.com/show/9319/amd-launches-carrizo-the-laptop-leap-of-efficiency-and-architecture-updates/6

It will be up to laptop manufacturers to use these, and that's where amd had problems in the past.. as far as i can see, intel has the monopoly. Maybe since this carrizo has an on-die arm core for security, more business grade laptops will be made with it.

[NikosD]

August 1, 2014



Outdated...

Nice catch, wrong link by me.

The only reason I can think of that, is that I had two tabs open and copied the wrong link.

The right link is here:
http://anandtech.com/show/9689/ecs-liva-core-review-a-fanless-core-m-minipc/5

[Yups]

Nice catch, wrong link by me.

The only reason I can think of that, is that I had two tabs open and copied the wrong link.

The right link is here:
http://anandtech.com/show/9689/ecs-liva-core-review-a-fanless-core-m-minipc/5



I checked the driver thread.....no Hybrid HEVC for Core M is intended: https://communities.intel.com/thread/59216

[CruNcher]

You can read a lot of technical information about Carrizo and its specs in the Anandtech article: http://www.anandtech.com/show/9319/amd-launches-carrizo-the-laptop-leap-of-efficiency-and-architecture-updates

15w TDP, up to 8.3 hours for h264 playback or 300 minutes for hevc 4k 60fps (with a 50wh battery), lots of power efficiency tweaks...

Page 6 has a lot of details about UVD6 :

http://www.anandtech.com/show/9319/amd-launches-carrizo-the-laptop-leap-of-efficiency-and-architecture-updates/6

It will be up to laptop manufacturers to use these, and that's where amd had problems in the past.. as far as i can see, intel has the monopoly. Maybe since this carrizo has an on-die arm core for security, more business grade laptops will be made with it.

Apu Silicon measured a total of 30 Watts for the Laptop he tested.

So they seem to more cram the Performance Gain not into a Power Saving but a Performance Gain especialy on the GPU side

https://www.youtube.com/watch?v=G4YT01OkYOo

Yeah but the Trustzone part seems only to be active in the business laptop part (for now) not the consumer the consumer is the FX-8800p the Business one is another Model.

But the thing is the Trustzone Capability is also in the Consumer version it's just sleeping same as the TPM part.

And you most probably can't change the OS on these Systems you will be locked to Windows 10, Installing Linux for example could become a major problem as Secureboot is Mandatory for Windows 10 upto Windows 8 it was still optional for the Manufactures to get the Windows Logo.

Now you will be forced to enter the new Signed Windows World and give up total control of the System like you do in the Apple world already.

[Xebika]

DXVA Checker v3.9.0 is out

Version 3.9.0 (2015/10/12)

Added the following alternative names of Decoder Device
VP8_VLD
VP9_VLD_Profile0

Version 3.8.0 (09/13/2015)

Now check 8K resolution for Decdoer Device and Processor Device
Changed "4KUHD" to "4K"
Fixed not detect "HEVC_VLD_Main10" when you select "H.265/HEVC" as format even if decoder support it
Minor changes and fixes

[P.J]

DXVA Checker v3.9.0 is out

Version 3.9.0 (2015/10/12)

Added the following alternative names of Decoder Device
VP8_VLD
VP9_VLD_Profile0

Version 3.8.0 (09/13/2015)

Now check 8K resolution for Decdoer Device and Processor Device
Changed "4KUHD" to "4K"
Fixed not detect "HEVC_VLD_Main10" when you select "H.265/HEVC" as format even if decoder support it
Minor changes and fixes

GTX960:
VP9_VLD_Profile0: DXVA2/D3D11, SD / HD / FHD / 4K

[nevcairiel]

Before you get all excited, Microsoft (or NVIDIA, or Intel) still needs to release a DXVA2 specification before anyone can implement a decoder for this.

Edit:
Actually, they did, and I just found it. Guess I have some work to do.

[huhn]

i highly doubt that VP9 gets important.

not part of UBD spec and the google VP9 encoder is nightmare slow.

[baii]

Chrome is pushing VP9 YouTube down everyone's throat. Say if Google buy Netflix or w.e.
i highly doubt that VP9 gets important.

not part of UBD spec and the google VP9 encoder is nightmare slow.


Sent from my 306SH

[huhn]

it's not like you have to play the VP9 version.

[wanezhiling]

Before you get all excited, Microsoft (or NVIDIA, or Intel) still needs to release a DXVA2 specification before anyone can implement a decoder for this.

Edit:
Actually, they did, and I just found it. Guess I have some work to do.
Wow, tha't great, can't wait. :p

ps: currently, only skylake and nvidia gm206 support vp9??

[nevcairiel]

Looks like it doesn't actually work yet on latest NVIDIA drivers. The specification is only 5 days old, so that may not be that big of a surprise. I should boot up my Haswell and see if that exposes it as well.

[NikosD]

I think you need Broadwell or better.

For Haswell is "empty"

[CruNcher]

Guess VP9 support comes with VP7 only like 10 Bit HEVC ?

Does AMD also supports it in the FX-8800p Carrizo UVD, i saw no demonstration of it in their marketing ?

Though neither Nvidia really seemed to advertise anything of it much.

Theoretically Nvidias Android TV Shield Console should also support it 10 bit HEVC + VP9 ?

Hmm those no named only IDs are the PP filter ?

http://i2.sendpic.org/i/e3/e3BCiamvZYeltcax0LMVWzvkMbG.png

[NikosD]

I checked the driver thread...

Using DXVA Checker v3.9.0, does Skylake's drivers provide VP8/VP9 support ?

[pacuro]

This is my first post on Doom9's. Reading here since few years and learning from the best.
Now have something to add and hope it helps.
I have GTX960 so confirm P.J's result regarding VP9 in DXVA Checker.
Additionaly asked my collegues for DXVA Checker 3.9.0 screenshots from i3 Broadwell NUC - HD5500 and i7 Skylake - HD530. Does it look like no VP9 hw support?

http://static.frazpc.pl/board/2015/10/c9cec425f38a5869.jpg

http://static.frazpc.pl/board/2015/10/07c206b1ad06527c.jpg

http://static.frazpc.pl/board/2015/10/b546c9a604df7b53.jpg

[nevcairiel]

The official (as in from Microsoft) VP8 and VP9 DXVA2 support was only officially released last week. Just because its not shown right now doesn't necessarily mean it never will be. Drivers may need to adapt to the final version of the specification first.

[pacuro]

... and another one collegue with Braswell NUC (Pentium N3700 - HD Graphics). Seems to be better, like GTX960.

http://i60.tinypic.com/nej7yq.jpg

[rbej]

No HEVC 10 bit support in Braswell??.

[CruNcher]

Anyone with VP7 (GTX960,GM206) Android Shield TV Console (X1), Maxwell Mobile starting with GTX 940M

can test this clip please on the latest driver

http://demo-uhd3d.com/fiche.php?cat=uhd&id=67



Jesus Samsungs NX1 samples where allready crazy but this is insane

http://demo-uhd3d.com/fiche.php?cat=uhd&id=112

[P.J]

This is my first post on Doom9's. Reading here since few years and learning from the best.
Now have something to add and hope it helps.
I have GTX960 so confirm P.J's result regarding VP9 in DXVA Checker.
Additionaly asked my collegues for DXVA Checker 3.9.0 screenshots from i3 Broadwell NUC - HD5500 and i7 Skylake - HD530. Does it look like no VP9 hw support?

Please test some of the previous clips with them :thanks:


Anyone with VP7 (GTX960,GM206) Android Shield TV Console (X1), Maxwell Mobile starting with GTX 940M

can test this clip please on the latest driver

http://demo-uhd3d.com/fiche.php?cat=uhd&id=67



Jesus Samsungs NX1 samples where allready crazy but this is insane

http://demo-uhd3d.com/fiche.php?cat=uhd&id=112

GTX960 has no issue with any HEVC clip :)

[nevcairiel]

Only 8-bit and only 100mbit? Thats not too bad. =p

If these are clips directly recorded with a camera without re-encoding, then they are usually not very "complex" clips, coding-wise, as cameras just encode with an extremely high bitrate, but not the complex coding tools, because its easier to build them that way.

[CruNcher]

Yeah Pardon it's Sonys XAVC not HEVC so overall the Samsung NX1 samples easily kill it in complexity

[NikosD]

This is my first post on Doom9's. Reading here since few years and learning from the best.
Now have something to add and hope it helps.


Welcome and thank you for your post.

Really helpful.

[NikosD]

I decided to upgrade my ADSL line (14 Mbps/ 1 Mbps) to FTTC+VDSL (50 Mbps / 5 Mbps) and I have to say that it's the first time that I managed to watch in realtime - using Web browsers - 4K youtube clips.

Now, I can upload easily even bigger test files :)

[ashlar42]

My HTPC has a GTX 660, under Windows 8.1 x64. I use Kodi as a media player, hence I believe I'm forced to use x86 versions of LAV Filters. Any point in evaluating this Lentoid sw decoder or is LAV better and capable of partially offloading h.265 decoding to GPU?
Thanks, I searched for Lentoid here but the latest result dated back to April.

Edit: forgot about my CPU, relevant for sw decoding, it's a Q9550S (http://ark.intel.com/products/40815/Intel-Core2-Quad-Processor-Q9550S-12M-Cache-2_83-GHz-1333-MHz-FSB).

[pacuro]

Please test some of the previous clips with them :thanks:

It will be really difficult, as these guys do not care a lot about quality and performace of playback and comfortable watching movies. They have sent me printscreens and it was painfull for them :D
What? MPC-HC? LAV? DXVA Checker? :D

Will try anyway, maybe when visiting them once.

[Yups]

Using DXVA Checker v3.9.0, does Skylake's drivers provide VP8/VP9 support ?


Looks like there is VP9 support on HD 530 (20.19.15.4300).


http://fs5.directupload.net/images/151025/4epk7yr4.jpg

[NikosD]

It's been discovered here in Doom9 by Wanezhiling - not for the first time I think - a serious performance hit for Intel's iGPUs when using NV12 video output (the default option for LAV Video and Microsoft's DXVA) compared to YV12.

You can read more here:
http://forum.doom9.org/showthread.php?p=1744279#post1744279

Now, according to those findings, I opened a thread to Intel's forum hoping for a fix.

The link/thread is here:
https://communities.intel.com/thread/88410

For anyone interested in Intel's iGPUs, you can open an account in Intel forums or use your existing one and add your voice with your experience about the subject, writing at the above Intel thread, in order to find a solution.

Nintendo Maniac 64 has already done that.

Intel has replied nothing till now.

I think we have to "push" them a little more!

Thanks!

[nussman]

Die you try to compare the Videoprocessing (Luma and chroma scaling, Deinterlacing)?
Maybe in YV12 you can not use the adaptiv GPU Deinterlacer (bob/weave Fallback) or the "good" GPU scaler (bilinear fallback).

[clsid]

Also disable all "enhancements". For example Skin Tone Enhancement and Contrast Enhancement.

[NikosD]

Die you try to compare the Videoprocessing (Luma and chroma scaling, Deinterlacing)?
Maybe in YV12 you can not use the adaptiv GPU Deinterlacer (bob/weave Fallback) or the "good" GPU scaler (bilinear fallback).
I think Intel should reply to all these.
What exactly are the differences of NV12 and YV12 regarding scaling algorithms or any other reason of slow performance.

Because for Nvidia and AMD, I haven't seen such differences.

Also disable all "enhancements". For example Skin Tone Enhancement and Contrast Enhancement.

I always do that.

Actually, it's the first thing I do for all my GPUs (Nvidia, AMD, Intel)

In Intel forums, I even suggested them to put everything off by default, but Intel thinks those "enhancements" are useful and tested by them.

[Kevin Gilmore]

Is it possible to 'pre-render' high-bitrate HEVC content
through Broadwell (or 10bit HEVC via Skylake:) Hybrid decoders such that I/one can then watch forthcoming/the future 4kBlurays (more high bit-rate HEVC/VP9 content) through my hybrid decoder solution in real time?

That is, Haswell/Broadwell/even SKYLAKE is only 'good' at best for real-time decoding at say ~18Mbps, given their respective formats (and shader faculties).

Put another way, or again......, I'd like to decode ~100Mbps or even say up to 600Mbps .265 files in advance, so that I can get more use than compressed youtube 4k from my Hybrid-solution decoders?
For instance, we'd desire to use a 'quasi-new Broadwell', and "new" Skylake chips to run my 4k TV or monitor at real-time rates, can we[B] 'pre-render/process' (I'd hope not transcode where Sky's capable!) Hbrfiles a la solution [for 'real-time' playback]?


It appears that Broadwell non-GT3e can't hybrid decode 10bit according to Intels' IGP driver.
People here have posted Broadwell Gt2 can indeed decode HEVC8/10 and VP9 though and I've seen references before-made that 5th gen 2+2 has been decoding VP9 (through YT HTML5 player) for this past year...

Some type of Processing or rendering anteceding playback would be a terrific H-br .265 'solution' for those without whole ff decoders (or say desktop gm206) until our next -particularly mobile- upgrade, (perhaps in Broadwell or Haswell/Skylake user case sometime in 2017/2018) etc.

Seeing as 6th/7th gen probably bring nothing else substantial to the table than Fixedfunction decoders - if Hybrid decoders could pre-process.... HBM, better OLED/DCPI3 =>4klaptop displays, and (noiseless) lower TDPs' are just around the corner.. Mentioned because they are very much relative to MadVR/LAV experiences and rendering..
6th/7th gen with accompanying GPUs' are going to be initially expensive circa release. I'm just trying to, "have the cake and eat it" now/today (as opposed to early thru late next year at best-case scenario in the mobile Form-factor) while prosaically saving a buck, with the hybrid decoders we already own.

Is preprocessing UHD-esque or High bitrate .265 possible for proper hybrid decoder playback? Appreciate and have empathy for those who read this WoTxt.

I've a Broadwell 5200U/940M if I can be helpful/run any tests..

[huhn]

not going to happen

first of all it would take hours to decode and the size of that decode file needs to be on a SSD.
a HDD is way to slow and to make things even worse the SSD needs to be huge too...

so forget it. HEVC hybrid decoders are kind of useless.

[vivan]

Yeah, 10bit 2160p@60 fps is 3840*2160*15*60 = 7 Gb/s. Which is around 50 GB/min.
Your only way is reencoding.

[leonccyiu]

it's not like you have to play the VP9 version.

That's true, but 4k60 videos are only encoded in vp9 and only show 1080p60 max in internet explorer. At the moment 4k60 videos are few, but as time passes they may become more common.

I also do notice that vp9 is in general less prone to artifacts and I think it is a superior compression algorithm. Whereas VP9 is used to reduce bit-rates for 1080p and below, it seems for 4k videos, many of them have higher bitrates, sometimes significantly so than h264.

I can't wait for full hardware acceleration to be working. Chrome seems to be worst at implementing it, even though on chrome://gpu everything is checked as working, when I use BBC iplayer or anything streaming h264 on html5 I can see cpu usage rise corresponding to video quality meaning hardware accelerated decoding is not working.

[NikosD]

I would call that a miracle, but as you know miracles have stopped happening the last few thousand years.

Latest Intel driver v4300 for Haswell Win 10 and better have "fixed" the performance drop of NV12 (!)

Now, using DXVA Checker and MPC-HC, I see no performance difference between NV12 and YV12 (!!)

BUT looking carefully at the picture of decoded clip, I see the same artifacts using NV12 like the ones using YV12.

Before v4300, the picture quality of NV12 was better.
Now it matches YV12, like performance.

This must be wrong, but it seems that Intel reacted insanely fast and changed the NV12 to YV12, but when ?

I reported that issue on October 25 and the drivers were ready by October 1-16 (maybe not closed completely, however)

Actions needed:
@Nintendo Maniac 64
@Wanezhiling
@leonccyiu

Can you test my findings ?

[wanezhiling]

There is no v4300 driver for win8.1 and so I can't test...

[foxyshadis]

BUT looking carefully at the picture of decoded clip, I see the same artifacts using NV12 like the ones using YV12.

Before v4300, the picture quality of NV12 was better.

What do you mean? Decoding bugs? Do you have any screenshots?

[NikosD]

I copy my comment from the other thread.

I was talking about the 4K VP9 clip I uploaded and some artifacts that were visible with YV12 and invisible with NV12.

Read here:

It is true that when the camera zooms in from 28 sec to 39 sec, the picture in windows mode looks like a little weird around the grass, like transparent water floating around.

In full screen (1920x1080), everything looks good.

[P.J]

Isn't the performance difference between those colorspaces because of their scaler method?

[nussman]

Latest Intel driver v4300 for Haswell Win 10 and better have "fixed" the performance drop of NV12 (!)

Now, using DXVA Checker and MPC-HC, I see no performance difference between NV12 and YV12 (!!)

BUT looking carefully at the picture of decoded clip, I see the same artifacts using NV12 like the ones using YV12.

Before v4300, the picture quality of NV12 was better.

I would call this a bug and not a fix .. :eek:
Disabling "good" processing for playback FPS benchmarks? Come on ...
I hope you dont "push" Intel to do something stupid ...

[NikosD]

It's not exactly playback benchmark.
It's plain playback, just measured in numbers.

The difference is huge, it's like you can play it in realtime or not (dropped frames or slow performance)

I've never told Intel to increase the performance by dropping the quality.

And between you and me, who am I to tell Intel anything ?

But we need more people to confirm or deny my findings.

[leonccyiu]

I would call that a miracle, but as you know miracles have stopped happening the last few thousand years.

Latest Intel driver v4300 for Haswell Win 10 and better have "fixed" the performance drop of NV12 (!)

Now, using DXVA Checker and MPC-HC, I see no performance difference between NV12 and YV12 (!!)

BUT looking carefully at the picture of decoded clip, I see the same artifacts using NV12 like the ones using YV12.

Before v4300, the picture quality of NV12 was better.
Now it matches YV12, like performance.

This must be wrong, but it seems that Intel reacted insanely fast and changed the NV12 to YV12, but when ?

I reported that issue on October 25 and the drivers were ready by October 1-16 (maybe not closed completely, however)

Actions needed:
@Nintendo Maniac 64
@Wanezhiling
@leonccyiu

Can you test my findings ?

Playing the LG Garden Demo and Panasonic Angkor Wat demo from demouhd3d, from the MADVR osd, it says that they are NV12 and they are playing back absolutely fine with no image artifacts whatsoever. The video looks a little washed out compared to what I remember in the showroom, but that's because of the poor contrast of the ips monitor I have.

The Youtube 4k vp9 videos are all YV12, I do not notice any major quality differences on my 1080p monitor.

[NikosD]

The test shall be done by disabling NV12 from LAV Video properties and then YV12 is used.

You could then compare, after you disable NV12, the performance and quality of YV12 compared to NV12 before and after v4300 drivers.

You shall use a 4K scaled to a smaller window video, in order to use the scaling algorithms.

That's it!

[nevcairiel]

And don't use madVR, it behaves entirely differently and is probably not affected by the same quirk. Only EVR.

[huhn]

just to make this more clear.

is only downscaling affected or is up scaling affected too?

[CruNcher]

I5-2400 Sandy Bridge 4 Cores Multithreaded Fluid Playback Results Lav Video (OpenHevc) CPU

8 Bit:

4K-Changhong-Furano-HEVC-HD.Club.mp4
4K-TCL-super_car-Part-1-@HD.Club.mp4
Demo Ultra-HD - TCL - 4K HEVC.mp4
LG_4K_View-the-Feeling.mp4
Hispasat_UHD_NAB-2014_HEVC.mkv
Beauty_3840x2160_120fps_420_8bit_HEVC_MP4.mp4

10 Bit:

Exodus_UHD_HDR_Exodus_draft.mp4
Samsung_Fashion_Show_Milan_UHD.ts (Very near of the Edge)
Samsung_SUHD_Tech_Story_Europe.ts
Samsung_UHD_Dubai.ts
UHD_ENT_Transformer_Quad.ts
UHD_PD_Tech_Story_Euro.ts
UHD_PQ_Dive_to_New_World.ts
UHD_PQ_Iceland_2.ts
UHD_PQ_The_City_Limits.ts


Close 10 bit fail:

Shut.Up.And.Play.Berlin.2014.ts (Very close 3 High Motion Scene glitching out, still counted as a fail)

Interesting Results:

Eutelsat.Demo.HEVC.ts <- They made a nice Artistic Chaotic Motion Test Sequence :)

[wanezhiling]

http://bluesky23.yukishigure.com/en/DXVAChecker.html
Version 3.10.0 (2015/11/07)
Added "MJPEG_VLD_PureVideo" as an alternative name of Decoder Device
Minor changes and fixes

http://i.imgur.com/ONJTooI.png

[NikosD]

http://bluesky23.yukishigure.com/en/DXVAChecker.html


By running the "Select Format" option and then selecting "Motion JPEG", is there any HW MJPEG decoder installed for Nvidia ?

Intel's driver installs "Intel Hardware M-JPEG decoder MFT" which is working for Haswell and it is a very fast HW decoder.

You could also check directly by using "Check Decoders" -> "Select Media File" and choosing a MJPEG video file.

If Nvidia installs a working HW MJPEG decoder, I could upload a 1080p and a 4K sample to test it.

[huhn]

http://abload.de/img/mjpegfwr9k.png

and it works.

edit: a short test shows the video engine isn't used for decoding. decompression JPEG or something like that is most likely a normal task for a GPU.

[NikosD]

and it works

OK.

Using Win 10 x64 - Core i7 4790 - iGPU HD 4600@1.5GHz - v4300 and Intel MJPEG HW decoder x64 with DXVA Checker v3.10.0 (in pure decoding mode) I got these results:
(CPU usage <10%, GPU usage 0%)

The first clip is 1080p50fps@105Mbps
Here:
https://www.sendspace.com/file/hfhw5r

For the first clip, I needed to put "Repeat Count" at least 5 times, because decoding was too fast.

Clip 1:
Intel MJPEG HW decoder x64 1584/1585/1584


The second clip is 2160p50fps@292Mbps
Here:
http://www.filedropper.com/crowdrun4knoqjpeg

Repeat count 5.

Clip 2:
Intel MJPEG HW decoder x64 401/402/401

[huhn]

looks like both ffdshoiw DXVA1 and NVDIA MJPEG doesn't support MKV. AVI works fine.

[P.J]

Please don't upload on filedropper =/

[huhn]

yeah that took some time to DL..

[huhn]

i remuxed the files and here are the results...
3770k 3.5 GHZ
960 GTX 1430 mhz
tested driver 358.87

decode mode.

UHD doesn't work.

1080p:

12% CPU usage
DXVA checker doesn't see the GPU usage
so looks like it is limited by the CPU.
200 - 210 -215
GPU-z says core 135 mhz 202.5 memory clock

about 55% GPU usage at that speed and the memory controller is at ~35%
so something is happening at the GPU not sure what...

this is either new or they don't care.

[NikosD]

Please don't upload on filedropper =/

It is the easiest way for me to upload > 300 MByte files.

Do you know any other file server working with a simple drag and drop for uploading > 300 MByte files ?



UHD doesn't work.



It should work according to DXVA Checker.

So, probably the 292 Mbps bitrate is forbidden for Nvidia's hybrid MJPEG decoder.


1080p:

12% CPU usage
DXVA checker doesn't see the GPU usage
so looks like it is limited by the CPU.
200 - 210 -215
GPU-z says core 135 mhz 202.5 memory clock

about 55% GPU usage at that speed and the memory controller is at ~35%
so something is happening at the GPU not sure what...

this is either new or they don't care.

The 1080p clip is still a 105 Mbps clip, so CPU is probably the bottleneck due to that bitrate.

Anyway, MJPEG is a QuickSync HW accelerated format for Intel, so is unbeatable by hybrid or CPU decoders.

[P.J]

i remuxed the files and here are the results...
3770k 3.5 GHZ
960 GTX 1430 mhz
tested driver 358.87

decode mode.

UHD doesn't work.

1080p:

12% CPU usage
DXVA checker doesn't see the GPU usage
so looks like it is limited by the CPU.
200 - 210 -215
GPU-z says core 135 mhz 202.5 memory clock

about 55% GPU usage at that speed and the memory controller is at ~35%
so something is happening at the GPU not sure what...

this is either new or they don't care.

Same here. LAV uses only one thread too.
I have never seen any MJPEG video...

[nevcairiel]

I have never seen any MJPEG video...

Some older home-video cameras recorded in that, but its otherwise not very common, indeed.

[P.J]

Some older home-video cameras recorded in that, but its otherwise not very common, indeed.

But CPU can handle 1080p MJPEG and even 4k well.

[NikosD]

Its a shame they didn't build an official DXVA2 specification for VP8/9 decoding, or released their proprietary way to do it (as it clearly goes through some DXVA2 device), oh well. Only being able to use MediaSDK and not implement it using existing DXVA2 decoders is sad.

After reading the above post I had sent an email to Intel - on August 19.

I got an official reply today that:

"VP9 dxva is already enabled at driver level.
VP8 dxva will be enabled post-Win10/2.
We(MSDK) don’t plan to provide DXVA MFT.
MSFT will provide them in OS at the next revision time frame."

Also, they told me that Haswell is not going to support VP8/VP9 in any way.

They are not aware of any ISV’s that are currently using those MSFT GUIDs, but I told them that you (the developer of LAV filters) are working on this.

Am I right ? What is your progress ?

[nevcairiel]

VP9 DXVA works on Braswell/Skylake with my code now. NVIDIA exposes the GUID as well, but thats still broken in recent drivers.
Didn't do anything for VP8 yet due to lack of testability. Once driver support arrives.

[NikosD]

VP9 DXVA works on Braswell/Skylake with my code now.


Is it hybrid for those two architectures ?

[nevcairiel]

From what I understand yes, Gen9 GPUs only have hybrid VP9 decode.

[NikosD]

OK, thanks. It makes sense.

But I didn't see VP9 DXVA in the nightlies.

Is there something holding you back of releasing the VP9 DXVA decoder ?

[Alecex]

So am I getting this right? The GTX970 has almost no positive effect on h.265 Decoding when compared to what the gtx960 is capable of?

Did anyone test the AMD Fiji Graphicscards or the Nano?

If the ATI r9 380 or the upcoming R9 380X (Antigua XT) supports h.265 4k 60p playback as well as the GTX960 does, then i would prefer the ATI beacuse its the better gameing card as well for the price.

The GTX 950 is supposed to be as good as the 960 for video but 60 usd cheaper so did someone confirm its h.265 capabilities?

[nevcairiel]

The GTX960 is superior to all the other cards you listed for HEVC/H.265 decoding, because all the other models either only implement 8-bit hybrid decoding (GTX970, AMD before Fiji), or lack 10-bit decoding (AMD Fiji).
On top of that, the GTX960 is the only of the lot that fully supports HDMI 2.0 and HDCP 2.2. GTX970 only has HDMI 2.0, but not HDCP 2.2, and AMD is limited to HDMI 1.4

[Alecex]

Thanks a lot!

I hope they release a GTX 960 TI at some point with 4gb and 256 bit interface...

its painfull to know how much more this Graphicscard could do with a proper interface.

Maybe one schould buy the gtx950 untill they release a 960 TI... instead of going directly for the 960. Or is there any limitations on the 950 video wise?

[nevcairiel]

There are GTX960 models with 4GB, which should be fine. The available memory bandwidth should be enough for any video usage.

There were rumors of a 960 Ti before, but that was in the beginning of this year, and since it hasn't shown up by now, its unlikely that it still will.
Pascal will hopefully (and quite likely) have the full decoder chip in its entire lineup, so that offers some new options then.

[Alecex]

Yes but with a 950 u save money if its as good as the 960 for Video, has anyone tested it? if not i could get a 950 first.

When is pascal expected to launch? I dont want to wate for Pascal my GTx460 is so loud and nerve-wrecking... plus it only does 1080p h.264 palyback/decoding. I wil get a pascal but untill then i need to change my gtx 460.

[CruNcher]

So am I getting this right? The GTX970 has almost no positive effect on h.265 Decoding when compared to what the gtx960 is capable of?

Did anyone test the AMD Fiji Graphicscards or the Nano?

If the ATI r9 380 or the upcoming R9 380X (Antigua XT) supports h.265 4k 60p playback as well as the GTX960 does, then i would prefer the ATI beacuse its the better gameing card as well for the price.

The GTX 950 is supposed to be as good as the 960 for video but 60 usd cheaper so did someone confirm its h.265 capabilities?

There is a effect that can be measured in a small energy saving with Nvidias Shader Decoder in GM204 but its relatively small compared to OpenHevc on the 4 Core I5-2400 and it can playback more complex streams and of course 10 bit that Nvidias Shader Decoder fails on not sure how this would look like if the Shader Decoder would operate on DX 12 (Win10) but on Windows 7 it's not really worth the hassle with the GTX 970 currently.

For low complex Streaming i would still overall prefer Nvidias Shader Decoder, because it is overall still more efficient compared to my I5-2400 (95W TDP) in Power Saving with such lower complex Online Bitstreams.

But the more closer you get to Broadcast Complexity the more it hits Nvidias Shader Decoder Efficiency vs the CPU alone and the CPU takes the lead here.

So the more efficient the CPU the weaker this will become and puff it's gone maybe already with the same region 4 Core Haswell CPU with Lower TDP on OpenHevc, this could be already the case.

I will still keep an eye if Nvidia is maybe able to improve their Shader Decoder furthe (if they should try it but why should they if they have VP8 ready ;) ) from what i measure i doubt this is possible to the degree it would need to conquer the 4 Core Sandy Bridge CPU OpenHevec Performance here anyways on the more Complex Broadcast Bitstreams ;)

So either GTX 960 or waiting for Pascal where the VP8 becomes the Standard Decoder for all Cards ;)

In either Way as you still want to play Games with good options on like 1080p 60 FPS neither the GTX 950/960 are sufficient today you need at least the GTX 970, especially for more complex GameWork (Compute) stuff like Hairworks or VXGI where the FPS can easily drop down to 1080p 30 FPS.

For normal WEB Online Bitstream Complexity your are fine even with the Nvidia Shader Decoder for stuff like for example 4K 10 bit Netflix you would need to fall back to the CPU or internal GPU if you have one that supports H.265 10 bit like Intels Gen9 Quicksync or AMD Carrizos UVD.

[NikosD]

So am I getting this right? The GTX970 has almost no positive effect on h.265 Decoding when compared to what the gtx960 is capable of?

Did anyone test the AMD Fiji Graphicscards or the Nano?

If the ATI r9 380 or the upcoming R9 380X (Antigua XT) supports h.265 4k 60p playback as well as the GTX960 does, then i would prefer the ATI beacuse its the better gameing card as well for the price.

The GTX 950 is supposed to be as good as the 960 for video but 60 usd cheaper so did someone confirm its h.265 capabilities?

I think nowadays you can't have it all.

Nvidia GTX 960 supports fully the H.265 (HEVC) codec, but lacks VP9 as all cards right now and it's not as powerful as R9 380 for gaming.

If you think about the future, meaning DX12 games, you should avoid buying a Nvidia card.
Maybe Pascal could be better, but right now Maxwell has the well known problems.

AMD on the other hand is behind Nvidia for HTPC since it doesn't support 10bit HEVC and HDMI 2.0, yet, although its cards are a lot better suited for DX12 gaming.

Intel has the same problems with AMD, lack of 10bit HEVC and you will definitely need a discrete card for gaming, although it has the faster decoder/encoder for H.264 and 8bit HEVC.

On the other hand, the content in HEVC is so little.

I think right now VP9 is more important than HEVC, since YouTube using Chrome uses that codec by default and sometimes explicitly (4K60fps clips)

If I were you and wanted desperately to buy something right now, I would buy an AMD card if DX12 gaming was my first priority or I would wait for next year's cards from both AMD and Nvidia that will probably have everything accelerated in HW (AVC, HEVC (8bit/10bit), VP9) and will certainly be faster in DX12 gaming than this generation.

[P.J]

I think right now VP9 is more important than HEVC

Satellites are broadcasting 4K in H.265 10bit format. 4K Blu-rays will hit the market soon.
Not all people have fast internet to watch 4K VP9, and it's useless for streaming 4K in low bitrate, could be much worse with fake upscaled videos too.
People can download the source video which has better quality than YouTube 4K.

And there's no proper DX12 game right now
http://www.anandtech.com/show/9740/directx-12-geforce-plus-radeon-mgpu-preview/4

[NikosD]

Satellites are broadcasting 4K in H.265 10bit format.


That's the case for very few people watching 4K H.265 DVB-S.


4K Blu-rays will hit the market soon.


There will be no 4K Blu-rays for 2015.
4K Blu-Ray will start sometime in 2016, but massively we don't really know.
Probably 2017 or even later


Not all people have fast internet to watch 4K VP9, and it's useless for streaming 4K in low bitrate, could be much worse with fake upscaled videos too.
People can download the source video which has better quality than YouTube 4K.


ALL YouTube videos and not only 4K are in VP9 format right now, when Chrome is used.
It's not that VP9 is used for 4K only.


And there's no proper DX12 game right now
http://www.anandtech.com/show/9740/directx-12-geforce-plus-radeon-mgpu-preview/4


As I said, DX12 gaming is a future (2016) thing too.

[P.J]

Recent CPUs can decode 1080p well but 4K.

[rodrigo081089]

Hi, i bought the gtx 960, i can reproduce 4k 8bit h264 videos ok, but when i tried 4k 10bit hevc i have a lot of dropped frames.

I use the MPC, could it the MPC reproduce HEVC ok?

[huhn]

make sure you are using DXVA not CUVID

[rodrigo081089]

make sure you are using DXVA not CUVID

If i use DXVA even the 1080p videos are unwatchable, i had a drop frame every second

[huhn]

than there is something wrong with your system. i can only guess but this is the wrong thread for this.

crappy cuvid can't do 10 bit.

[rodrigo081089]

than there is something wrong with your system. i can only guess but this is the wrong thread for this.

crappy cuvid can't do 10 bit.

I reinstalled the drivers again and now i could make work DXVA. When i open a 720p or 1080p it says H/W decoding, but when i open a 4k video it´s only software decoding.

Maybe someone know how to change that?


Thanks and sorry for the offtopic

[wanezhiling]

MPC? MPC-HC or MPC-BE? Install their latest version from here: MPC-HC (https://nightly.mpc-hc.org/), MPC-BE (http://mpc-be.org/builds/)
And then reset all settings (You know how to do that right?:D), ok now 4k h/w decoding is surely back. :)

[rodrigo081089]

MPC? MPC-HC or MPC-BE? Install their latest version from here: MPC-HC (https://nightly.mpc-hc.org/), MPC-BE (http://mpc-be.org/builds/)
And then reset all settings (You know how to do that right?:D), ok now 4k h/w decoding is surely back. :)

Great, it was that. If i use Madvr i cant get h265 decoding by hardware.


Thank you, now i can sleep :D

[v0lt]

Nvidia GTX 950
http://i.imgur.com/pmjtdT5.png

Beauty_3840x2160_120fps_420_8bit_HEVC_MP4.mp4
decode: http://i.imgur.com/Es7Hfog.png
playback 1280x720: http://i.imgur.com/E6Kq1ko.png

It is NVIDIA CUVID mode. Tests for DXVA (native) mode here (http://forum.doom9.org/showthread.php?p=1746626#post1746626).

[wanezhiling]

@v0lt
Try UHD_PQ_Lovely_Swiss.60fps.ts which is 10bit

[NikosD]

Nvidia GTX 950

Beauty_3840x2160_120fps_420_8bit_HEVC_MP4.mp4


The first results on January by nevcairiel were slower using GTX 960.

Decode & Playback (1080p scale): 125fps

This 15% increase of GTX 950 (149-145 fps) could be drivers, better DXVA decoder or faster VP clock engine (don't think so)

[nevcairiel]

I just ran it again, and I get similar results now, 145 fps in average. Maybe they did improve drivers.
Except, I get the expected 0-2% CPU usage from a full DXVA2 decoder. Whats up with v0lt's CPU usage there?

PS:
Note that I used a debug version of the decoder, since I was too lazy to setup everything for proper benchmarking. 1-2% differences in performance are expected due to that. :)

[v0lt]

In the previous test (http://forum.doom9.org/showthread.php?p=1746607#post1746607) it seems to have been inadvertently selected NVIDIA CUVID.
Here are the results for the DXVA (native).

Nvidia GTX 950
DXVA Checker (http://i.imgur.com/qJChsHX.png)

DXVA (native) mode.

8-bit
Beauty_3840x2160_120fps_420_8bit_HEVC_MP4.mp4 (link (http://ultravideo.cs.tut.fi/video/Beauty_3840x2160_120fps_420_8bit_HEVC_MP4.mp4))
decode: 151 fps (http://i.imgur.com/TsxLyLO.png)
playback 1280x720: 151 fps (http://i.imgur.com/4JOY9Oq.png)

10-bit
UHD_PQ_Lovely_Swiss.ts (link (http://www.demo-uhd3d.com/fiche.php?cat=uhd&id=26))
decode: 121 fps (http://i.imgur.com/jcb5Bdi.png)
playback 1280x720: 121 fps (http://i.imgur.com/qJChsHX.png)

[CruNcher]

Not from the Decoder but testing the current state of the Hevc Encoder of GM204 on the GTX 970 (trying to reach highest speed)

NVEnc 1.10 (x64), using NVENC API v5.0
OS Version Windows 7 (x64)
CPU Intel Core i5-2400 @ 3.10GHz [TB: 3.20GHz] (4C/4T)
GPU GeForce GTX 970 (13 EU) @ 1266 MHz (358.78)
Input Buffers CUDA, 16 frames
Input Info y4m (yv12) -> nv12 [AVX], 1920x1080, 25/1 fps
Output Info H.265/HEVC main
1920x1080p 1:1 25.000fps (25/1fps)
Rate Control VBR
Bitrate 4000 kbps (Max: 17500 kbps)
Initial QP I:20 P:23 B:25
VBV buffer size auto
GOP length 250 frames
B frames 0 frames
Ref frames 3 frames
MV Quality Q-pel

4 Mbits 1080p (max) = 174-180 fps

[Alecex]

So the GTX 950 is just as fast for Video decoding/playback. Nice!
lets see what i II get maybe just a 950 untill new NVidia Generation

[CruNcher]

If you take it as a option into account for Media Playback it would make more sense then to go with Nvidias Android Shield TV for a bit more

[huhn]

and you are totally doomed if you want to play a file that can't be hardware accelerated.

[CruNcher]

We know what the VP8 is capable of by now through all our tests pretty much same should apply for the Android Shield TV if you have other knowledge please share :)

[huhn]

you can't play 10 bit AVC and a lot of other codecs the list is really really long, VP9 doesn't work yet and maybe is going to be hybrid which will not work on a "CPU" like the shield CPU.

there is a ref frame cap on DXVA decoding don't ask me for the number...

[NikosD]

there is a ref frame cap on DXVA decoding don't ask me for the number...

The Ref frames restrictions have all gone at least since 2009.

ATI was the last one to support Ref 16 with UVD2.2, not from the beginning but with a driver update (the HW was already capable)

[CruNcher]

you can't play 10 bit AVC and a lot of other codecs the list is really really long, VP9 doesn't work yet and maybe is going to be hybrid which will not work on a "CPU" like the shield CPU.

there is a ref frame cap on DXVA decoding don't ask me for the number...

Somehow i doubt this as it would mean no Netflix 4K support ?

[sneaker_ger]

Netflix uses HEVC for UltraHD, not 10 bit AVC or VP9.

[huhn]

netflix is HEVC isn't it?

[CruNcher]

Netflix uses HEVC for UltraHD, not 10 bit AVC or VP9.

Ah sorry auto replaced it with 10 bit HEVC ;)

ok so no 10 Bit AVC not really important for any content service platform though running currently in business terms, but yeah sad anyways for most non avg user content.

Most Anime Fansub viewer will get lost so for Nvidia, most probably they don't care much about that target group of users anyways in their calculation and decisions around VPx.

And in reality it could be enough resources left for Web Content VP9 with Hybdrid Decoding for the Shield Android TV so if it doesn't run already nicely in their Shield browser it's surely a matter of time VP9 services like Youtube will work.



Another thing i think i found another way of how the DXVA Native problem shows up for HEVC and the Lav Video Decoder on GM204 it shows way differently in its effect then in MPC-BEs OSD spikes only but it causes real visual issues in DXVA Checkers playback that look like Frame Ordering issues when they appear at certain spots at playback, interesting :)

http://i2.sendpic.org/t/8L/8LIKFDTBjHfBRArxmAZ3drMZSN9.jpg (http://sendpic.org/view/2/i/5ApYYLED2B5FvFFfmlk2IyHXi4m.png)
http://i2.sendpic.org/t/Ad/Ad3MdjNdzFxHpkCFdhCCQa6lQDB.jpg (http://sendpic.org/view/2/i/bWJqH3qK8jvyhwNusl972nhoIs.png)

As you can see it looks pretty stable from the Data but at playback it wasn't it showed Frame Ordering issues (jumpy frames forward/backward) at several places.

here in MPC-BE you wont see these Frame Ordering issues only the spikes DXVA Native is causing @ playback with Lav Video Decoder and recognize some small slowdown in this test case.

http://sendpic.org/view/2/i/o5NmFsOU8aEiG714hE5ZlgTJcR4.png




So something is different with DXVA Checker :)

DXVA Native = Frame Jumps
DXVA Copyback = No Frame Jumps
NVCuvid = Frame Jumps
None (OpenHevc) = No Frame Jumps

that is really strange that Cuvid also shows Frame Jumps and only DXVA Copy Back and Software show none.

[rodrigo081089]

The Man In the High Castle 3840x2160 H264 10bit doesnt work with DXVA

[sneaker_ger]

H.264(=AVC) 10 bit hardware/DXVA decoders do not exist in the consumer world.

[NikosD]

Impressive, but it is for 2H of 2016.

I think all big 3 (AMD, Intel, Nvidia) will have those features in 2016 - I'm not sure about VP9 10bit. I haven't seen any clip.

[wanezhiling]

AMD, I don't think so.

[Yups]

http://i.imgur.com/iPxdPTl.jpg

Kaby Lake supports fixed function HEVC 10bit and VP9 10bit decoding.



It's a Gen 9.5 graphics by the way. It might support HDMI 2.0 as well.

[huhn]

AMD, I don't think so.

i'm pretty sure they will support it. there is no reason to skip it and they don't relabel again...

in the past they where the first with bitstreaming and 5.1 PCM.

[Alecex]

VP9 would be good for chrome youtube but they might set it "on" with the drivers at some point...

I now have a palit gtx960 4GB Jetstream because its so cool and quiet.
But There is a Problem. When I play 4k Youtube Videos in Mozilla Firefox a get micro stutter. My CPU stays at very low usage while the 960 is getting the load.
Thats the video
https://www.youtube.com/watch?v=TmDKbUrSYxQ
It happens every 1 second. It seems to be like the PullDown Problem we had in the early days of 24p BluRay when wie didnt have tru 24p TVs

[CruNcher]

No issues here on Firefox 45 internal Player for the H.264 stream with the HTML5 player but the quality of the Moving Watter Surface encoding for a 4K result is really awefull :(
You really would want to avoid something like this for 4K

[Alecex]

sure its awefull but at least it does play hardware accelerated and my computer does not run hot... my gtx 460 did not accelerate anything above 1080p and it also got loud with any acelerated video...

how do i see if its the html 5 player or the flash? edit... foudn it

[Yups]

i'm pretty sure they will support it. there is no reason to skip it and they don't relabel again...

in the past they where the first with bitstreaming and 5.1 PCM.


You can't be sure without any reliable leak. As for APUs first chance it will happen is sometime in 2017 with a Zen based APU.

[daWsOn_s]

Was anyone able to get a real time playback on i5 intel ivy bridge for 4K HEVC?

I'm using the i5-3570k overclocked at 4Ghz and with mpc-hc and lav I can't get a decent playback.

Thanks.

[Alecex]

Do the Skylake i5 6200U and so on with the HD Graphics 520 have real HW VP9 h.265 Decoder like the Desktop Skylakes? If so has any one benched them?

[P.J]

Was anyone able to get a real time playback on i5 intel ivy bridge for 4K HEVC?

I'm using the i5-3570k overclocked at 4Ghz and with mpc-hc and lav I can't get a decent playback.

Thanks.

Which video exactly?

[NikosD]

New beta drivers from Intel v4326 - they cover even Win 10 Nov Update (10586)
https://downloadcenter.intel.com/download/25541/Intel-Beta-Graphics-Driver-for-Windows-7-8-1-10-15-40-

igfxHK module crashed during installation and after restart too - I disabled "hotkeys" from the Control Panel.

The drivers support MediaSDK API v1.17 and probably VP8_VLD.

Can someone using latest Win 10 Nov Update with Broadwell or Skylake report with DXVA Checker if VP8_VLD is present and if there are any new VP8/VP9 decoders (using Check Decoders -> Select format from DXVA Checker) ?

[CruNcher]

GPU GeForce GTX 970 (13 EU) @ 1266 MHz (359.00) (GM204) WDDM 1.1

H.265 Encoder PCIE 2.0 Transcoding (CPU Decoding/GPU Encoding)

4 Mbits 1080p (max 17.5 Mbits) = 181 FPS

Not sure if anything changed on the Hybdrid Decoding side doesn't look like it though from the First tests 4K Decoding Performance seems still slightly behind 8 bit OpenHevc 64Bit on 4 CPU Cores (Sandy Bridge TB: 3.20GHz no HT) :(

I guess Nvidia stopped Optimizing the Maxwell Shader Decoder so most probably we wont see any more gains here for VP6, i at least hoped they would reach similiar OpenHEVC Performance then Intel 4 Core Sandy Bridge IPC alone but they don't reach it for the same Extreme Bitstream complexity Levels at least not on the GTX 970 13 EUs.

Maybe though they reach it under Windows 8.1 (WDDM 1.3) or 10 (WDDM 2.0) can't say that yet :)

[wanezhiling]

http://xhevc.com/en/downloads/downloadCenter.jsp
6.The 10-bit encoded HEVC video can be correctly decoded (but still rendered in 8-bit for output).
Now on my laptop 4k hevc 10-bit 60fps can be played smoothly, with FFmpeg its just horrible.

[CruNcher]

WOW what for a CPU ? also Atemes Broadcast Streams complexity (Samsung Korea) ?

[sneaker_ger]

I don't see much of a difference between Lentoid 32/64 and LAV 64. As usual, only LAV 32 pales in comparison. Mainly interesting for those stuck to 32 bit software like DVB Viewer.

Tested on Core i7-860 (SSE4.2, no AVX), GraphStudioNext, Null Renderer:

UHD_PQ_Lovely_Swiss.ts (10 bit)
lav 64: 47.7 fps
lentoid 64: 45.1 fps
lentoid 32: 43.8 fps
lav 32: 19.0 fps

Ducks.Take.Off.2160p.h.265-Zion.mkv (8 bit)
lav 64: 39.9 fps
lentoid 64: 39.3 fps
lentoid 32: 38.0 fps
lav 32: 20.1 fps

Sad to see LAV's speed hasn't really improved in over a year, at least not for non-AVX CPUs. Seems like the OpenHEVC optimizations still haven't been ported to ffmpeg?

[P.J]

http://xhevc.com/en/downloads/downloadCenter.jsp

Now on my laptop 4k hevc 10-bit 60fps can be played smoothly, with FFmpeg its just horrible.

It doesn't even support DXVA2 :confused:

[NikosD]

New updated CPU results for LAV, Lentoid and Microsoft's H.265 decoder

System:
Win 10 x64 (10586) - Core i7-4790 - iGPU HD4600@1.5GHz - v.4326

Decoders:
LAV x64/x86 0.67.3
Lentoid x64/x86 2.1.0.0
Microsoft HEVC decoder build 10586

Tools:
DXVA Checker 3.10.0 in Benchmark Decode mode

Splitter:
LAV Splitter x64/x86 0.67.3



1.ProRes-1080p@30fps-2Mbps



Lentoid x64 320/398/460 CPU 84%

LAV x64 309/396/467 CPU 78%

Lentoid x86 280/380/453 CPU 84%

MS HEVC x64 280/362/410 CPU 87%

MS HEVC x86 165/222/275 CPU 51%

LAV x86 126/166/227 CPU 87%




2.Elephants Dream-1080p@24fps-1.7Mbps



MS HEVC x64 139/360/490 CPU 85%

Lentoid x64 120/352/594 CPU 63%

LAV x64 123/328/565 CPU 55%

Lentoid x86 113/302/490 CPU 65%

MS HEVC x86 77/260/415 CPU 52%

LAV x86 66/145/360 CPU 59%



3.Big Buck Bunny-1080p@60fps-2Mbps



MS HEVC x64 280/410/490 CPU 86%

LAV x64 231/400/560 CPU 52%

Lentoid x64 169/386/530 CPU 57%

Lentoid x86 193/340/490 CPU 61%

MS HEVC x86 178/333/442 CPU 53%

LAV x86 71/162/365 CPU 53%



4. Girls -1080p60 fps -11Mbps


MS HEVC x64 223/290/348 CPU 91%

Lentoid x64 199/284/355 CPU 81%

LAV x64 187/269/335 CPU 78%

Lentoid x86 190/260/330 CPU 82%

MS HEVC x86 126/191/252 CPU 52%

LAV x86 94/133/190 CPU 81%


5.Beauty-2160p@30fps-12.3Mbps



Lentoid x64 69/85/91 CPU 67%

LAV x64 71/84/89 CPU 69%

MS HEVC x64 71/80/83 CPU 77%

Lentoid x86 63/79/86 CPU 75%

MS HEVC x86 52/59/63 CPU 47%

LAV x86 32/44/48 CPU 88%



6.Fitness-2160p@30fps-8Mbps



LAV x64 93/107/128 CPU 81%

Lentoid x64 66/99/127 CPU 89%

Lentoid x86 66/90/118 CPU 90%

MS HEVC x64 70/89/110 CPU 88%

MS HEVC x86 52/67/92 CPU 51%

LAV x86 37/58/105 CPU 91%



7.Ducks-2160p@50fps-4Mbps



LAV x64 103/113/116 CPU 86%

Lentoid x64 85/100/105 CPU 78%

MS HEVC x64 91/98/100 CPU 90%

Lentoid x86 73/89/92 CPU 89%

MS HEVC x86 73/78/81 CPU 55%

LAV x86 37/46/48 CPU 91%


8.Gravity - 2048x858@24fps-5Mbps-10bit



Lentoid x64 127/196/353 CPU 76%

LAV x64 119/187/393 CPU 60%

Lentoid x86 118/179/354 CPU 79%

MS HEVC x64 99/150/307 CPU 93%

LAV x86 74/120/348 CPU 66%

MS HEVC x86 58/92/308 CPU 51%


Comments:


1) Lentoid added 10 bit decoding with 8bit output (I420).
MS HEVC decoders 10bit decoding uses 8bit output too (NV12)

2) Lentoid x86 keeps running faster than LAV x86 by a huge margin and it's faster than MS HEVC x86 too.

3) Lentoid x64 is very close to LAV x64 and generally faster but there are strong fluctuations in the results, so sometimes it's difficult to measure both decoders.

4) The real surprise is MS HEVC x64 which is faster than anything, but only for 1080p.
In the first clip, it has some problems, it even crashes DXVA Checker when GPU acceleration is used.

But for all the other 1080p clips is definitely faster.
On the other hand, MS HEVC decoders are slow on 2160p clips and especially 10bit clips

It seems - most of the times - that whenever each decoder manages to utilize more CPU usage, it ends first in the specific clip.

[Manni]

Hi everyone,

At the moment I have an AMD HD7870 which struggles with many UHD/4K files, especially 10bits ones.

I had ruled out upgrading to the AMD Fury/Fury X due to the lack of HDMI 2.0 (18Gb/s) and HDCP 2.2 support, but this looks like it's going to be solved by the upcoming Club3D displayport 1.2 to HDMI 2.0 active adapter (http://club-3d.com/index.php/products/reader.en/product/displayport-12-to-hdmi-20-uhd-active-adapter.html). I could wait for Arctic Islands, but waiting until at least Q2 2016 isn't a great prospect.

So my question is: would a Fury (I don't think I'll stretch to the Fury X as this is mainly for better video with MPC-BE/LAV/MAdVR) significantly improve this?

Reading the thread I'm not sure if the Fury supports hardware acceleration for x265 (and VP9, although that's not that much of an issue for me right now).

Does anyone know what the status is with the new Crimson 15.11 driver, which should support the Club3D adapter?

My CPU (see sig) is not the latest and greatest, so I need as much GPU acceleration I can get.

I tried to find answers but I didn't find any solid info about the state of h/w or hybrid decoding for the new AMD GPUs.

Thanks!

[chros]

http://xhevc.com/en/downloads/downloadCenter.jsp

Now on my laptop 4k hevc 10-bit 60fps can be played smoothly, with FFmpeg its just horrible.
And what about LAV video x64 if only enable the 4:2:0 output? (I can't play back 10bit with 10bit output on my system (too slow for that), but I can in 8 bit.)

[sneaker_ger]

4:2:0 is chroma sub-sampling, not bitdepth. There is both 4:2:0 8 bit and 4:2:0 10 bit.

For LAV any kind of chroma resampling or 10 bit -> 8 bit dithering means additional computation, not faster processing. Either LAV or the renderer has to do this (in most cases). If your system struggles with the 10 bit output it might have to do with your renderer.

[nevcairiel]

Having LAV convert 10 to 8-bit is nearly free from a performance standpoint, so using that instead of 10-bit could result in an overall performance increase.

[NikosD]

I updated my post with MS HEVC decoder results.

What a surprise!

MS managed to reach the top once again but only for 1080p clips

I clearly remember talking about the subject, saying that the development of HEVC decoder by Microsoft hasn't ended yet, even if official Win 10 build 10240 had released.

A lot of people disagreed with me, telling me that the official Win 10 release meant the end of HEVC decoder development.

Well, we can now see I think, that HEVC decoding development still goes on for Microsoft reaching the top once again.

Unfortunately, 2160p HEVC decoding is slow using MS HEVC decoders and especially in 10bit clips.

MS should put more effort on that.

[NikosD]

At the moment I have an AMD HD7870 which struggles with many UHD/4K files, especially 10bits ones.


So my question is: would a Fury (I don't think I'll stretch to the Fury X as this is mainly for better video with MPC-BE/LAV/MAdVR) significantly improve this?

Reading the thread I'm not sure if the Fury supports hardware acceleration for x265 (and VP9, although that's not that much of an issue for me right now).

Does anyone know what the status is with the new Crimson 15.11 driver, which should support the Club3D adapter?



Decoding UHD/4K H.264 & H.265 would be no problem with a Fury card as long as those clips are 8bit.
Because Fury has real HW decoder for 8bit H.264 & H.265 codecs.

4K 10bit HEVC clips could be a problem if the bandwidth is high, like the upcoming BluRay 4K.

Regarding VP9, I don't think Fury has hybrid support, but AMD could add it to the driver in the future.

I don't know if Crimson has changed that.

Your CPU (Core i7 3770K) is fast enough for any low bandwidth clip (H.265 and VP9)

[Manni]

Decoding UHD/4K H.264 & H.265 would be no problem with a Fury card as long as those clips are 8bit.
Because Fury has real HW decoder for 8bit H.264 & H.265 codecs.

4K 10bit HEVC clips could be a problem if the bandwidth is high, like the upcoming BluRay 4K.

Regarding VP9, I don't think Fury has hybrid support, but AMD could add it to the driver in the future.

I don't know if Crimson has changed that.

Your CPU (Core i7 3770K) is fast enough for any low bandwidth clip (H.265 and VP9)

Thanks a lot for this.

The demo clips I'd like to be able to test are all under 50mb/s. They are detected as BT2020 by MadVR but it's just the container, they are most likely rec709 or P3.

UHD Bluray, we'll see, it causes other issues like P3 gamut, HDR, etc. Plus it might take a while before the new AACS is cracked or PDVD supports it. I'll probably use a standalone player for a while anyway just for these.

I guess if a Fury helps me to upscale x264 bluray to UHD with full MadVR quality in 10bits 4:4:4 and to play a few HEVC demos I'll be happy until I can get an Arctic Island towards the end of 2016.

[envagyok]

At the first post, what mean the three numbers end of lines?

1.ProRes-1080p@30fps-2Mbps
https://www.sendspace.com/file/mo8exg


LAV x64 i7-4790 158/230/251

LAV x86 i7-4790 121/151/188

LAV x64 i7-4790 (DXVA) 83/134/157

LAV x86 i7-4790 (DXVA) 84/133/158

Strongene x86 i7-4790 (OpenCL) 83/112/146

Strongene x86 i7-4790 97/109/111

LAV x64 i5-4200M 76/98/114

LAV x64 i5-4200M (DXVA) 60/84/96

LAV x64 Core 2 Quad 45/52/54

[NikosD]

The three numbers always mean:

Minimum/average/maximum number of frames per second (fps)

The more the the faster, obviously.

[Nullack]

In trying to get an idea of UHD bluray decoding for when discs are purchasable, I've so far found a sample with HEVC Main 10 L52 2160P60 using a constant bit rate of 51.5 Mbps. Using my GTX 960 with nightly MPC-HC x64, nightly Lav Filters x64, in DXVA 2 native mode on the EVR-CP renderer, all is good with only minor paint times maximums around 25ms against the 60 fps cadence of 16 ms. The great majority of frames are being done in around 5-6 ms, which is heaps of headroom below the 16 ms 60 fps cadence. The statistics aside to the eye it all looks nice and smooth. Is there any other sample at a higher bit rate that would be closer to test what the UHD bluray discs might be like?

[NikosD]

Is there any other sample at a higher bit rate that would be closer to test what the UHD bluray discs might be like?

This is a 10bit HEVC 2160p50 fps test clip at 127Mbps:
https://www.sendspace.com/file/ietm9n

This a HDR 10bit HEVC 2160p60 fps clip at 45Mbps:
http://demo-uhd3d.com/files/uhd4k/NEO_Fire_HDR_HEVC_2160p_5994_Main10_45-100VBR.mp4

Give them a try and tell us your experience.

[chros]

Is there any other sample at a higher bit rate that would be closer to test what the UHD bluray discs might be like?
But won't movies/tv shows be (stay) 24p?

[Nullack]

This is a 10bit HEVC 2160p50 fps test clip at 127Mbps:
https://www.sendspace.com/file/ietm9n

This a HDR 10bit HEVC 2160p60 fps clip at 45Mbps:
http://demo-uhd3d.com/files/uhd4k/NEO_Fire_HDR_HEVC_2160p_5994_Main10_45-100VBR.mp4

Give them a try and tell us your experience.

Thanks. That short clip of the runners, under the same config I described previously that 127 Mbps bitrate makes no difference in the telemetry as my 60 FPS L52 MAIN 10 sample. Meaning, the same video engine utilisation, the same GPU core utilisation, the same gpu memory usage. Also the EVR-CP jitter and paint time stats are around the same too. So basically that setup gobbles up the video of the runners race no sweat.

The BT.2020 took more fiddling to test because I had to switch my colour mode in my UHD 4K display and that required removing the HDMI, changing the setting, putting it back then changing the desktop settings too. Once all the colour was configured OK. I had to drop down from 60 hz to be able to do the BT.2020 HDR over HDMI 2 - my display only has HDMI for UHD. I played it back under the same config and it made no difference to the EVR stats or the telemetry of how the GPU was loaded. So it seems that with HEVC 4K playback, the HDR BT.2020 basically comes for free in terms of measuring its impact on the GPU and the renderer stats in comparison to "normal" L52 MAIN 10 HEVC.

[nevcairiel]

HDR or BT.2020 doesn't make decoding a sample more complex, its just a different way to interpret the image data afterwards.

[NikosD]

Thanks. That short clip of the runners, under the same config I described previously that 127 Mbps bitrate makes no difference in the telemetry as my 60 FPS L52 MAIN 10 sample. Meaning, the same video engine utilisation, the same GPU core utilisation, the same gpu memory usage. Also the EVR-CP jitter and paint time stats are around the same too. So basically that setup gobbles up the video of the runners race no sweat.


Maybe the same clip at 300Mbps could give a hard time for GTX 960:
http://www.filedropper.com/crowdrun2160p50main10300mbps

[foxyshadis]

300Mbps is way beyond UHD bluray specs though. There the maximum is 128Mbps, and that applies to every 1-second window (I haven't analyzed the 127Mbps test to see if it complies yet).

[huhn]

still "working". first keyframe is broken and even my CPU can do about 18-40 FPS does this even use cabac?

decode speed is down to 75 - 84 - 88

everything over 150 mbit is kind of unrealistic.

edit: it's about 14 FPS with the CPU so everything is normal just a wrong number in EVR.

[nevcairiel]

does this even use cabac?

HEVC doesn't have an alternative coding mode like H.264 did.

[huhn]

i see.

it just looked like an easy to decode stream on the first look. (for 300 mbit)

[NikosD]

decode speed is down to 75 - 84 - 88

everything over 150 mbit is kind of unrealistic.



Pushing a little more the envelope, I uploaded the same clip at 500Mbps.
Just to find out the bandwidth limit of 10bit HEVC decoder of GTX 960.

http://www.filedropper.com/crowdrun2160p50main10500mbps

I have encoded up to 1000Mbps the same clip in 100Mbps step, back on February with x265 encoder, but I don't think that 500Mbps are feasible by 960.

[huhn]

it's dropping the first keyframe already so it can't really handle the 300 mbit correctly.

[NikosD]

Besides the first keyframe, your minimum fps is 75, so it can handle the rest of the clip.

[huhn]

55 - 67 - 73

you get just an corrupted image until the next keyframe.

with madVR and "delay playback until all queues are full" playback is smooth.

[NikosD]

55 - 67 - 73

you get just an corrupted image until the next keyframe.

with madVR and "delay playback until all queues are full" playback is smooth.

600Mbps:
http://www.filedropper.com/crowdrun2160p50main10600mbps

[huhn]

54 -61 -67

[Nullack]

Maybe the same clip at 300Mbps could give a hard time for GTX 960:
http://www.filedropper.com/crowdrun2160p50main10300mbps

Mate that 300 Mbps clip using native LAV and my usual setup EVR-CP, there is visible distortion in colour and the trees/cloud to the left are artifacing too. In at about 5 seconds of playback it pickups up a bit with the right colours but its not smooth. The 127 Mbps runners are OK, the 300 pushes it over. MPC-HC nightly doesnt crash but the playback is not correct.

[huhn]

like i said a keyaframe is dropped.

[NikosD]

Last chance to saturate the 10bit HEVC decoder - 1023Mbps
http://www.filedropper.com/crowdrun2160p50main101000mbps

[Nullack]

like i said a keyaframe is dropped.

Yes but that wouldnt explain the stuttering would it, I believe a gtx 960 was designed to do the UHD BD spec max bitrate ok, and at 300 Mbps its overloaded

[huhn]

no my 960 isn't stuttering what so ever.

[huhn]

Last chance to saturate the 10bit HEVC decoder - 1023Mbps
http://www.filedropper.com/crowdrun2160p50main101000mbps

41 - 43 - 47

that's too much.

[v0lt]

Last chance to saturate the 10bit HEVC decoder - 1023Mbps
Artifacts and freezes on the GTX 950.

[Xebika]

DXVA Checker v3.11.0 is released.

Changelog:
Version 3.11.0 (2015/12/06)
Added support to enable/disable DXVA2 for Microsoft H265 Video Decoder MFT
Minor fix

Download:
http://bluesky23.yukishigure.com/en/DXVAChecker.html

DSF/MFT Viewer v1.4.2 is released.

Changelog:
Version 1.4.2 (12/06/2015)
Minor fix

Download:
http://bluesky23.yukishigure.com/en/DsfMftViewer.html

[P.J]

Thanks, just noticed that Microsoft H265 Video Decoder MFT can't use GPU in D3D11 mode and it's very very slow in CPU mode!

Edit: LAV doesn't use P010 in CPU mode/DXVA2 copy-back =/

[nevcairiel]

Edit: LAV doesn't use P010 in CPU mode/DXVA2 copy-back =/

Actually, it does, but EVR does not accept P010 in those modes. When you use benchmarking mode without a renderer, its used.

[NikosD]

For those who didn't understand it from one of my previous posts, Microsoft has added (at last) DXVA2 HEVC decoder in Win 10 November Update (build 10586)

Up to now, only D3D11 HEVC acceleration was available from Microsoft.

MS MFT DXVA2 HEVC decoder works for Intel and it is a little faster than LAV DXVA2 HEVC using my iGPU and sometimes a lot faster.

[NikosD]

Looks like there is VP9 support on HD 530 (20.19.15.4300).


The new LAV nightly build 0.67.23 from here http://files.1f0.de/lavf/nightly/ has added VP9 DXVA2 acceleration.

It would be interesting to test it on your Skylake HD 530 and make a comparison with VP9 CPU Skylake decoding using this clip:
https://www.sendspace.com/file/pr4m1c

[rbej]

I've just pushed VP9 DXVA2 support to LAV, which will be available in the next nightly build tomorrow.
Note that there is some caveats to know about:

- Its limited to 4:2:0 8-bit (Profile 0), like most DXVA2 accelerators
- Its been tested on Intel Braswell only. Skylake will probably also work. Older Intel systems may not.
- While NVIDIA pretends to support VP9 DXVA2 on the GTX960, it does not actually work (black screen). I'm somewhat confident that its a driver problem, since it fails in rather specific ways that it otherwise shouldn't.
- When using DXVA2-Native and playing an unsupported file (4:2:2/4:4.4 or > 8-bits), you may get a black screen, since fallback to software may not catch the unsupported format early enough (Copy-Back does not have this issue, since it can fallback to software at any time)

Due to the limited amount of hardware that can actually make use of this, testing was also limited, so keep that in mind.
Its also disabled by default, due to all this.

Braswell VP9 decoding benchmark.

http://i.imgur.com/hNLszAW.png

VP9 4k 25fps file

http://www.artbyphil.com/temp/phfx_4KHD_VP9TestFootage.zip

HEVC 4k Braswell decoding benchmark

http://i.imgur.com/awoQMSy.png

[NikosD]

Can you test this file with CPU and GPU decoding of your Braswell ?
Which Braswell is exactly ?

https://www.sendspace.com/file/pr4m1c

Also, generally run every benchmark at least 3 times, not just a single one.

Thanks!

[Aleksoid1978]

Intel(and Nvidia - but can't test right now) have Hybrid decoder VP9 DXVA2 - that's why so slow and High CPU usage.

[NikosD]

Skylake with HD 530 should be a lot faster though, even with a hybrid VP9 decoder.

[nevcairiel]

On my NUC5PPYH (ie. Pentium N3700)

Decode Benchmark: 25/31/50, at 40% CPU usage

The clip would maybe just barely run, but probably not quite fluid.

[NikosD]

But is it faster than pure CPU Pentium N3700 decoding ?

[nevcairiel]

But is it faster than pure CPU Pentium N3700 decoding ?

CPU decoding benchmark (32-bit):
19/31/81 - 84% CPU

64-bit results were almost identical.

The VP9 software decoder in FFmpeg is already optimized quite well, so it fares OK even on this CPU.
Right now, it doesn't seem like the hybrid decoders offer a big advantage - at least compared to FFmpeg software decoding.

Maybe Skylake will show some more interesting results.

[NikosD]

You are right about Braswell, but still you have almost the same speed decoding by dropping CPU usage from 84% to 40%.

That's a gain too.

[NikosD]

I had ruled out upgrading to the AMD Fury/Fury X due to the lack of HDMI 2.0 (18Gb/s) and HDCP 2.2 support, but this looks like it's going to be solved by the upcoming Club3D displayport 1.2 to HDMI 2.0 active adapter (http://club-3d.com/index.php/products/reader.en/product/displayport-12-to-hdmi-20-uhd-active-adapter.html). I could wait for Arctic Islands, but waiting until at least Q2 2016 isn't a great prospect.


Good news for AMD fans.
According to the official statements of AMD, the upcoming GPUs on 2016 will support all the latest and greatest protocols for HTPC, like HDMI 2.0a (HDMI 2.0 + HDR), DisplayPort 1.3 and HDCP 2.2 for HDR movies.

Also they will add FreeSync support over HDMI.

For more details you can read here:
http://anandtech.com/show/9836/amd-unveils-2016-vistech-roadmap

The only thing left for clarification is the HW accelerated support of HEVC 10bit and VP9 (8bit/10bit)

[Manni]

Good news for AMD fans.
According to the official statements of AMD, the upcoming GPUs on 2016 will support all the latest and greatest protocols for HTPC, like HDMI 2.0a (HDMI 2.0 + HDR), DisplayPort 1.3 and HDCP 2.2 for HDR movies.

Also they will add FreeSync support over HDMI.

For more details you can read here:
http://anandtech.com/show/9836/amd-unveils-2016-vistech-roadmap

The only thing left for clarification is the HW accelerated support of HEVC 10bit and VP9 (8bit/10bit)

Good news indeed. Note that Club3D is coming up with an active DisplayPort 1.2 to HDMI 2.0 (full 18Gb/s bandwidth) with HDCP 2.2 in a few weeks which brings most of this to the current gen with the Crimson drivers. The only thing missing is HDR support (and of course DisplayPort 1.3), which should only come with Arctic Island.

http://club-3d.com/index.php/products/reader.en/product/displayport-12-to-hdmi-20-uhd-active-adapter.html

I'm getting a Club3D adapter for my HD7870 and I'm waiting for Arctic Islands to upgrade the GPU.

As you know, I tested a Fury last night and the lack of hybrid acceleration for Main10 is a dealbreaker for me. It only helps with x264 upscaling (I don't really need it, I'm happy with 1080p for bluray) and x265 8bits (little real content is ever going to be 8bits, as both UHD Bluray and UHDTV need 10bits).

As AMD is supporting the Club3D adapter with Crimson, there is a tiny chance that they'll add the HDMI 2.0a profile as the hardware (both on the adapter and the GPU) can do it, but it's very unlikely to happen.

[nevcairiel]

Honestly it would've been a shame if the next gen GPU didn't offer these things. Same with HEVC 10-bit, if Arctic Island doesn't offer that for some reason, it would be sad.
VP9 might or might not happen. Considering DXVA2 only specifies hardware acceleration for 8-bit, its probably unlikely to see 10-bit support though, but maybe someone surprises us.

[NikosD]

@Manni

PowerDVD offers hybrid acceleration of HEVC 10bit via OpenCL as nevcairiel said, for Fury.

I remember testing 8bit HEVC OpenCL hybrid decoding on my Haswell and it worked fine.

Maybe I have to search for latest version of Cyberlink's HEVC decoder to test 10bit HEVC hybrid on Haswell too.

@nevcairiel

I think it's easy for MS to add 10bit VP9 profile into DXVA2, now that has already done the first crucial step of supporting VP9 8bit.

[Manni]

Honestly it would've been a shame if the next gen GPU didn't offer these things. Same with HEVC 10-bit, if Arctic Island doesn't offer that for some reason, it would be sad.
VP9 might or might not happen. Considering DXVA2 only specifies hardware acceleration for 8-bit, its probably unlikely to see 10-bit support though, but maybe someone surprises us.

Agree 100%. The lack of HDMI 2.0 18Gb/s is the reason why I didn't upgrade to the Fury earlier (only the Club3D adapter made me consider it again) and the lack of GPU power is what stopped me to upgrade to the 960, as it's at best a side step power-wise compared to my 7870.

I can't believe that there is still no decently powerful GPU offering HDMI 2.0a with HEVC Main10 acceleration (at least hybrid) and more power than a 3 year old GPU.

I don't really care about VP9, but if they don't provide HEVC Main10 acceleration, it looks like I'm going to keep my HD7870 for a while :)

[Manni]

@Manni
PowerDVD offers hybrid acceleration of HEVC 10bit via OpenCL as nevcairiel said, for Fury.

I remember testing 8bit HEVC OpenCL hybrid decoding on my Haswell and it worked fine.

Maybe I have to search for latest version of Cyberlink's HEVC decoder to test 10bit HEVC hybrid on Haswell too.


Thanks, I missed Nevcairiel's post on HEVC Main10 via OpenCL for PDVD.

That explains the difference and tallies with my testing :)

[nevcairiel]

@nevcairiel

I think it's easy for MS to add 10bit VP9 profile into DXVA2, now that has already done the first crucial step of supporting VP9 8bit.

It would be the first time ever that they extended a codec beyond its "first" profile, and after releasing the spec at that. Which is why I think its unlikely - but like I said, maybe they'll surprise us!
It would probably require someone other than YouTube starting to pick up VP9 first, so it gets more wide-spread usage.

[NikosD]

Didn't MS do the same thing with HEVC ?

I thought that they implemented 8bit first and later did the Main10.

[nevcairiel]

Didn't MS do the same thing with HEVC ?

I thought that they implemented 8bit first and later did the Main10.

No, the HEVC DXVA2 spec contained both 8 and 10 right from the start.
HEVC was smart, their first version of the spec contained both 8 and 10 bit, so anyone claiming to support HEVC Version 1 needs to support both. :D

[P.J]

PowerDVD's 10-bit HEVC decoder is quite slower than LAV, both in CPU mode.


For those who didn't understand it from one of my previous posts, Microsoft has added (at last) DXVA2 HEVC decoder in Win 10 November Update (build 10586)

Up to now, only D3D11 HEVC acceleration was available from Microsoft.

MS MFT DXVA2 HEVC decoder works for Intel and it is a little faster than LAV DXVA2 HEVC using my iGPU and sometimes a lot faster.

Then why can't GTX960 use Microsoft H265 Video Decoder MFT in D3D11 mode in DXVA Checker?

[NikosD]

Then why can't GTX960 use Microsoft H265 Video Decoder MFT in D3D11 mode in DXVA Checker?

Are you asking for this reply ?

http://forum.doom9.org/showthread.php?p=1732881#post1732881

[P.J]

Are you asking for this reply ?

http://forum.doom9.org/showthread.php?p=1732881#post1732881

Ah right, thanks.

It's impossible to connect to EVR with a DXVA11 decoder, like the MS MFT HEVC decoder, so the DXVA Checker could never become red and could never use that decoder to HW accelerate HEVC content because it uses EVR renderer.

So, DXVA Checker will always fall back to SW using MS MFT HEVC decoder in "Playback mode" or "Playback Benchmark" mode as it uses EVR.


Edit: No way to add Microsoft H265 Video Decoder DXVA2 to MPC-HC? :confused:

[NikosD]

MS decoder is a MediaFoundation decoder, so it's not compatible directly with MPC-HC, but LAV has already done that with MS WMV decoder.

IIRC, nevcairiel has said that if MS MFT H.265 decoder is faster than LAV, he could probably add it to LAV with a DS wrapper on top of MFT decoder.

[wanezhiling]

I would call that a miracle, but as you know miracles have stopped happening the last few thousand years.

Latest Intel driver v4300 for Haswell Win 10 and better have "fixed" the performance drop of NV12 (!)

Now, using DXVA Checker and MPC-HC, I see no performance difference between NV12 and YV12 (!!)

Actions needed:
@Nintendo Maniac 64
@Wanezhiling
@leonccyiu

Can you test my findings ?
Tried 10.18.14.4332 (http://forum.doom9.org/showpost.php?p=1749903&postcount=2660) with my HD4600 (Win 8.1) and I can confirm that the performance drop of NV12 was really fixed, indeed.:)

[NikosD]

Can you compare the quality with YV12 ?

It looks like exactly the same to me, that's why they (NV12 & YV12) have the same speed.
Because the have the same quality.

Try playback with scaling, e.g a 4K clip to a 720p window.

The old NV12 was a lot slower but with obviously better quality too.

[wanezhiling]

Yes, I can see the difference, not very visible here though.
I suspect old NV12 was still using some 'hidden' driver features (even you disable all video releted settings in driver control panel) while YV12/new NV12 are not.

[NikosD]

Right.

I always disable everything in the control panel, but the old NV12 was better than the new one in the same configuration (with everything disabled).

For me in scaling it's obvious.

[Aleksoid1978]

Result VP9 DXVA on GTX 960(latest drivers):

Decoder: LAV Video Decoder
[1]
Decoder Device: VP9_VLD_Profile0
Frames: 2880
Time: 23,532 sec
FPS: 122,388 [79-141] fps
CPU Usage: 2 [0-3] %
GPU Usage: -
[2]
Decoder Device: VP9_VLD_Profile0
Frames: 2880
Time: 22,480 sec
FPS: 128,114 [106-141] fps
CPU Usage: 2 [0-3] %
GPU Usage: -
[3]
Decoder Device: VP9_VLD_Profile0
Frames: 2880
Time: 22,530 sec
FPS: 127,828 [106-143] fps
CPU Usage: 2 [0-3] %
GPU Usage: -


using phfx_4KHD_VP9TestFootage.webm file.