DXVA benchmarks - Avivo vs PureVideo vs Clear Video

[NikosD]

UPDATE 23/09/2010 - Added Vista results with CPU & GPU usage, DXVA1/2 results, MPEG2 results - Removed DivX (incompatible with DXVAchecker - breaks MediaFoundation) - PostProc OFF (Catalyst Video settings & codecs)

A thread for anyone interested in DXVA benchmarking.

The tool of DXVA benchmarks is DXVA checker v2.3.0.0 x86 version, available here:
http://bluesky23.hp.infoseek.co.jp/en/index.html#DXVAChecker

Systems:
Win Vista SP2 x86 - C2D@2.83 GHz - Radeon 5750 (UVD 2.2) - Catalyst 10.9a
Win 7 x64 - C2D@2.83 GHz - Radeon 5750 (UVD 2.2) - Catalyst 10.9a

UVD 2.2 codec support: (according to ATI)

MPEG2, WMV3, VC-1, H.264

There is no support for MPEG2_VLD yet. Only hardware assisted support for MPEG2 with DXVA1/2 versions

For WMV3, VC-1, H.264 there is support of up to 1088p. For VC-1, H.264 there is support of 1088p60fps

Cut-off point is 1088. Max results below for carefully chosen samples of each codec.

Catalyst 10.4 and later introduced full support of H.264 L5.1 up to 4K x 2K.
Never activated and confirmed. Only support of 1088p L5.1 H.264 up to now (Catalyst 10.9a)

DXVA decoders tested: (x86)

1) FFDShow DXVA rev 3576
2) MPC-Video decoder v1.4.2561 standalone filters
3) ArcSoft Video decoder v2.27.319.108
4) Cyberlink Video decoder v1.0.0.1505 (PowerDVD 10)
5) Microsoft DTV-DVD video decoder DS (DirectShow)
6) Microsoft H264 Video Decoder MFT (Media Foundation)
7) Microsoft MPEG-2 Video Decoder (DirectShow)
8) AMD Fast (DnD) decoder (Media Foundation)

Video Formats:
VC-1, H.264, WMV3, MPEG2

All samples can be found here:

ftp://helpedia.com/pub/multimedia/x264/testvideos/


VC-1

1) 1080p30fps - 12Mbps

2) 1080p60fps - 40Mbps


H.264

1) L4.1-1080p30fps ReF 2 - 27Mbps

2) L5.1-1080p24fps ReF 16 - 34Mbps

3) AVCHD - L4.2-1080p60fps ReF 2 - 30Mbps

4) L5.1-2304p24fps ReF 1 - 6Mbps



WMV3

1) 1080p24fps-10Mbps


MPEG2

1) 1080p25fps-60Mbps


Instructions:

1) Download DXVA Checker and sample clip

2) Open DXVA Checker and click the big horizontal button Check DirectShow/MediaFoundation Decoders...

3) Select the downloaded clip

4) Left click the decoders shown by the program - the ones with red colour - and you will see a menu. Go down to benchmark option and select EVR for DXVA2 or VMR7 for DXVA1

5) It opens a new window. Select preview off. Write down the Min/Avg/Max fps


Results: (Min/Avg/Max), CPU & GPU usage average only

Vista x86

MPEG2


DXVA1

(MPEG2_C) ArcSoft 90/96/101 CPU: 43% GPU: 0%
(MPEG2_A) Microsoft DS 85/93/100 CPU: 44% GPU: 0%
(MPEG2_C) Cyberlink 77/86/95 CPU: 44% GPU: 0%

DXVA2

(MPEG2_IDCT) Microsoft DS 88/94/100 CPU: 44% GPU: 0%
(MPEG2_IDCT) ArcSoft 89/89/90 CPU: 42% GPU: 0%
(MPEG2_IDCT) Cyberlink 76/85/94 CPU: 44% GPU: 0%


WMV3

DXVA2 only


AMD MFT: 67/70/91 CPU: 3% GPU: 0% (crashes DXVAChecker after test)
ArcSoft: 55/70/93 CPU: 3% GPU: 0%


VC-1


First clip:

DXVA1

FFDshow: 67/70/91 CPU: 4% GPU: 4%
MPC-HC: 67/70/91 CPU: 4% GPU: 4%

DXVA2

AMD MFT: 68/73/92 CPU: 4% GPU: 0% (crashes DXVAChecker after test)
ArcSoft: 66/70/97 CPU: 4% GPU: 0%
FFDshow: 67/70/85 CPU: 3% GPU: 0%
MPC-HC: 67/70/85 CPU: 3% GPU: 0%

Second clip:

DXVA1

FFDshow: 65/68/88 CPU: 5% GPU: 4% (Artifacts during normal playback)
MPC-HC: 65/68/88 CPU: 5% GPU: 4% (Artifacts during normal playback)


DXVA2

AMD MFT: 66/68/95 CPU: 4% GPU: 0% (crashes DXVAChecker after test)
ArcSoft: 59/68/95 CPU: 5% GPU: 0%
FFDshow: 57/68/95 CPU: 3% GPU: 1% (Artifacts during normal playback)
MPC-HC: 57/68/95 CPU: 3% GPU: 1% (Artifacts during normal playback)


H.264

DXVA2 only


First clip:

MPC-HC: 50/59/89 CPU: 3% GPU: 0%
ArcSoft: 50/59/89 CPU: 3% GPU: 0%
FFDshow: 50/59/87 CPU: 3% GPU: 0%
Cyberlink: 50/59/76 CPU: 3% GPU: 0%

Second clip:

MPC-HC: 45/50/75 CPU: 4% GPU: 0%
ArcSoft: Not working
FFDshow: Not working
Cyberlink: Not working

MPC-HC: 332/332/332 - CPU: 34% GPU: 0% - Enabled unstable experimental feature (driver bug?)


Third clip: None of the following decoders is capable of constant realtime 60fps decoding

MPC-HC: 52/56/81 CPU: 2% GPU: 0%
FFDshow: 52/56/80 CPU: 2% GPU: 0%
Cyberlink: 52/56/70 CPU: 2% GPU: 0%
ArcSoft: 48/56/70 CPU: 2% GPU: 0%


Fourth clip:

No working DXVA decoders


Win 7 x64

CPU usage disabled in Win 7 by DXVAChecker.

MPEG2


DXVA1

(MPEG2_C) ArcSoft 90/95/100 GPU: 0%
(MPEG2_A) Microsoft DS 83/85/85 GPU: 0%
(MPEG2_C) Cyberlink 77/85/94 GPU: 0%

DXVA2

(MPEG2_IDCT) ArcSoft 88/89/90 GPU: 0%
(MPEG2_IDCT) Cyberlink 77/85/93 GPU: 0%
(MPEG2_IDCT) Microsoft DS 82/82/83 GPU: 0%


WMV3

DXVA2 only


AMD MFT: 68/72/115 GPU: 0% (crashes DXVAChecker after test)
ArcSoft: 66/70/89 GPU: 0%


VC-1

DXVA2 only for Win 7


First clip:


AMD MFT: 68/72/96 GPU: 0% (crashes DXVAChecker after test)
ArcSoft: 68/72/92 GPU: 0%
FFDshow: 68/71/85 GPU: 0%
MPC-HC: 68/71/85 GPU: 0%

Second clip:


AMD MFT: 66/68/93 GPU: 0% (crashes DXVAChecker after test)
ArcSoft: 66/68/93 GPU: 0%
FFDshow: 65/68/87 GPU: 0% (Artifacts during normal playback)
MPC-HC: 65/68/87 GPU: 0% (Artifacts during normal playback)


H.264

DXVA2 only


First clip:

Microsoft MFT: 52/72/107 GPU: 0%
Microsoft DS: 50/60/86 GPU: 0%
MPC-HC: 50/60/83 GPU: 0%
ArcSoft: 50/60/83 GPU: 0%
FFDshow: 50/60/83 GPU: 0%
Cyberlink: 50/60/76 GPU: 0%

Second clip:

MPC-HC: 45/51/88 GPU: 0%
Microsft DS: 34/51/92 GPU: 0%
Cyberlink: Not working
ArcSoft: Not working
FFDshow: Not working


MPC-HC: 391/391/391 GPU: 0% - Enabled unstable experimental feature (driver bug?)
Microsft DS: 210/240/270 GPU: 0% - Enabled unstable experimental feature (driver bug?)

Third clip: The DS decoders are broken. DXVAChecker v2.4 will probably fix this problem

Microsoft MFT: 54/63/106 GPU: 0%


Fourth clip:

No working DXVA decoders

Feel free to add comments and results

[renq]

It's a L5.1 - 1080p24fps with 16 ReF H.264 clip
ATI, Nvidia, Intel cards
Since this is a L5.1 clip, ATI HD2000 and HD3000 series cards are unable to play this using DXVA.

Just FYI.

Good going for 4500mhd

[NikosD]

MS DTV-DVD Video Decoder

Video Preview off
Video Preview on

Min/Avg/Max

Intel 4500MHD
60/70/108
55/68/100


9600M GT
32/42/62
0/42/62

For once, I feel proud of that Intel graphics card.

Yeah, I didn't see it coming.

What is the CPU usage during benchmark ?

[renq]

An idea, why not also test 4K/2160p/UHD?

Clips can be found HERE (http://forum.doom9.org/showthread.php?t=155857&highlight=2160p) or HERE (http://xhmikosr.1f0.de/index.php?folder=c2FtcGxlcw==) or everywhere on the Interwebs

[NikosD]

Since this is a L5.1 clip, ATI HD2000 and HD3000 series cards are unable to play this using DXVA.

Just FYI.

Good going for 4500mhd

Sure, I know.
I did it on purpose to push things as far as it gets.
But because I want everyone to be happy, a new sample.

Actually it's a movie trailer and it's a L4.0 High profile H.264 -1080p24 fps ReF 3 clip

Get it from here:
http://rapidshare.com/files/417906037/Paranormal_Activity.mkv.html

My results (always preview off)

Min/Avg/Max

DivX DXVA
44/54/57

FFDshow DXVA
44/55/67

Microsoft DTV-DVD
48/55/59

It looks like some kind of driver limitation holds back the hardware.

This clip is a lot easier than the first one.

[NikosD]

An idea, why not also test 4K/2160p/UHD?

Clips can be found HERE (http://forum.doom9.org/showthread.php?t=155857&highlight=2160p) or HERE (http://xhmikosr.1f0.de/index.php?folder=c2FtcGxlcw==) or everywhere on the Interwebs

Look my post here for the reply:
http://forum.doom9.org/showpost.php?p=1432028&postcount=30

[nevcairiel]

All with the MS DTV-DVD Decoder, on the initial 5.1 Sample
----------------------------------------
i7 860 with a NVIDIA GTX 260

min/avg/max

0/42/61

----------------------------------------
Phenom II x4 955 - NVIDIA GTX 460

min/avg/max

20/54/89

Turning Preview off fixes the really low minimal value, must be some bug somewhere, the other values don't change when turning it off.

[mariner]

Greetings NikosD and namaiki.

1. Can you also test the Media Foundation decoders as well? You may need to use a container like avi.
2. Testing the Panasonic TM700 1080@60p clips.

Many thanks and best regards.

[NikosD]

Did you have a look in wikipedia? It reports acceleration of AVC as 'Full,' but to be honest I have no idea if that is correct or not.

From Intel
Intel® G41 and G43 Express Chipsets provide partial H.264 decode hardware acceleration.
Intel® G41 and G43 Express Chipsets provide partial VC-1 decode hardware acceleration.

[namaiki]

From Intel
Intel® G41 and G43 Express Chipsets provide partial H.264 decode hardware acceleration.
Intel® G41 and G43 Express Chipsets provide partial VC-1 decode hardware acceleration.

My GPU is the 4500MHD, not the X4500.
4500MHD is on the GM45 chipset.

[NikosD]

Greetings NikosD and namaiki.

1. Can you also test the Media Foundation decoders as well? You may need to use a container like avi.
2. Testing the Panasonic TM700 1080@60p clips.

Many thanks and best regards.

WMP11 in Vista and WMP12 in Win 7, are the only players I know using MF decoders.

I have done some testing with H.264 in .mov container and found out that MF decoder is a little faster than DS decoder

[NikosD]

My GPU is the 4500MHD, not the X4500.
4500MHD is on the GM45 chipset.

Intel and Wikipedia say same things about X4500 and 4500MHD chipsets. X4500 partial, 4500MHD full.

Your tests with 4500MHD do not verify the above info.

Full acceleration with 45% CPU usage ? Funny definition of full.

[mariner]

WMP11 in Vista and WMP12 in Win 7, are the only players I know using MF decoders.

I have done some testing with H.264 in .mov container and found out that MF decoder is a little faster than DS decoder

I mean using DXVAChecker to benchmark/test the MF decoders.

Both AMD and Nvidia MF decoder are installed with the graphics drivers package.

[nevcairiel]

"ModeH264_VLD_NoFGT_ClearVideo" is just the name of the VLD profile on Intel cards. All their DXVA profiles have special GUIDs (and therefor names), and also require slightly different interaction with the decoders, which caused some trouble in ffdshow/mpc-hc dxva before (still does with some encodes).

14% CPU usage on a 2.4ghz isn't too bad. On my i7 860 i got about 2% usage, however on my Phenom II 955, i got like 10% as well. (Granted, Windows sees 8 cores on the i7 which could lower the average value)

Why the CPU usage is so significantly higher when using the Intel decoder instead of the NVIDIA decoder .. who knows. Might be something the decoder is doing, hard to say without any comparison.

[NikosD]

I mean using DXVAChecker to benchmark/test the MF decoders.

Both AMD and Nvidia MF decoder are installed with the graphics drivers package.


The .mov clip mentioned before is here:
http://rapidshare.com/files/416586142/1.Twinpeaks1080p30fpsRef2-27Mbps.mov.html

My results:

FFDShow DXVA: 47.17 fps
Microsoft DS: 47,75 fps
Microsoft MFT: 68,07 fps

DXVAChecker does not give Min/Avg/Max, has a problem with .mov container.
Just the final average number .

I haven't found any other sample.
If you have one, provide it.

[CKHSJ]

Clear Video is better, you can watch them in a good feeling, just enjoy your clear video.

[NikosD]

Is there anyone out there with an Atom processor (GMA 500/600) or Core iX with intergrated graphics (GMA HD) to give us his results ?

Or Nvidia VP3. That would be interesting, too

Thanks

[tetsuo55]

This thread started off great, lets keep it civilized.

I'm highly interested to see results from the mpc-hc standalone video decoder.

[NikosD]

This thread started off great, lets keep it civilized.

I'm highly interested to see results from the mpc-hc standalone video decoder.

MPC-HC Video Decoder v1.4.2499 x86 standalone filter

L5.1 clip results:

Min/Avg/Max
44/49/69

Exactly in the middle of DivX and Microsoft DTV-DVD decoders.

[tetsuo55]

Its very difficult to to compare these benchmarks, can be a bit more scientific about it?

-Always use the files and application version linked in the start post
-Run every benchmark with preview OFF
-Reboot before benchmarking
-Run 3 benchmarks, and present the average results
-Run a test on both the easy and the difficult sample
-Try ALL dxva decoders you can get your hands on, some examples: Ffdshow, mpc-hc-standalone, microsoft(both dshow and mediafoundation), cyberlink, divx, etc..
-Try all 3 codecs (h264, VC1, MPEG2)
-put all your results in one post, so its easy to compare results

:thanks:

Maybe this information will help us improve the open source decoders so they become pack leaders.

[NikosD]

Unfortunately not a lot more work is required by me.
Because:

1) MPEG2_VLD is not supported by ATI drivers or ATI hardware - who knows?.
That means that no DXVA MPEG2 appears when I run DXVAchecker.
Only partial acceleration by Microsoft DXVA1-VMR (ModelMPEG_A) and Microsoft DXVA2-EVR (ModelMPEG2_IDCT). That's all

2) VC1_VLD is supported by ATI.
But only FFDShow DXVA and MPC-Video decoder appear with exactly the same performance. Less than 1FPS diference. No Microsoft. At all

3) H.264
The results given, follow the rules you described.

There are no other DXVA H.264 decoders besides those given. Cyberlink PowerDVD 10 does not expose the DS decoder for public use.

I owe you to make a results table easily understood.

[mariush]

Well I'm not sure what I'm screwing up but on my Windows 2003 with Radeon 4850 I can't get it play with DXVA.
DXVA Checker reports: http://savedonthe.net/image/672/dxva.png

Tried with the posted l5.1 file and a 720x576 file that used to work perfectly when I encoded it a year or so ago. DXVA Checker only detects CoreAVC 2.0 and Divx H.264 Decoder and both say Unsupported. The ffdshow tryouts and mpc standalone filter don't appear (though I can't use either of them in MPC-HC either, maybe it's because I didn't restart the system...)

With the internal MPC-HC codec, the l5.1 file won't play in dxva and the 720x576 plays with artifacts (since a few driver versions it's screwed up)

[NikosD]

Catalyst version ?

[NikosD]

Unlike for my GeForce 9600M GT, GPU-z doesn't report GPU usage for the Intel GPU.

I meant GPU usage in DXVAChecker, it's the same as GPU load in GPU-z

[NikosD]

DXVA Checker doesn't report it for the Intel GPU.

Understood.
Could you report GPU load by GPU-z?

[NikosD]

As said before, GPU-z does not report GPU load for the Intel GPU. It only reports core and memory. Unless you meant something else.

My bad.
You mentioned Geforce and I thought you meant VPU usage at your first post, because we are talking about DXVA.
All clear now.
Intel has no metrics for GPU and VPU

[NikosD]

FYI, ATI UVD has no metric for Video Engine Load (VPU usage)
ATI card has only GPU load, reported by both DXVAChecker and GPU-z.

During H.264 and VC-1 DXVA benchmarking, GPU load is a dead 0%, as it should be.

[noee]

FYI, using the AMD GPU clock tool, shows UVD usage for my HD2600XT.

[NikosD]

FYI, using the AMD GPU clock tool, shows UVD usage for my HD2600XT.

Impossible is nothing for you.
Three simple questions:

1) How did you benchmark L5.1 H.264 clip in UVD ? You need UVD2.x

2) How did you manage to get numbers of GPU usage by AMD GPU Clock tool ? It only shows the status of UVD.

3) What are your numbers of GPU load ?

[noee]

I was only replying that the AMD GPU Clock tool would give usage for UVD and that perhaps it would work for the later-gen cards (I know my card will not do DXVA L5.1 h.264 with it's UVD rev.)?

[NikosD]

I have just updated my benchmarks at first post.

Feel free to add comments and your results.

[tetsuo55]

Thanks for testing
I think we can make something really good out of this.

We have 4 questions that we want to see answered.

1. For any given hardware decoder (uvd, vp, etc..) which codecs does it claim to support
2. For each codec, does the decoder support it up to its theoretical limits, and if not what is the cut-off point
3. For the cut-off point, what is the performance in FPS
4. Which driver version is required for each unique hardware decoder to get the best results

Maybe we can get all 4 points into 1 table and put that on wikipedia or something?
That should level the playing field considerably and would make it far easier to rewrite (and keep updated) the dxva detection code in mpc-hc and to recommend dxva hardware to others.

It will also be a push for hardware developers to improve their drivers and hardware, so as to score higher on the tests.

Targets would be:
Support ALL features of ALL codecs @ 60fps for the highest possible resolution.

There are more VLD codecs than have been tested here, namely mpeg1, and mpeg4-asp(divx) and RMVB(but i dont know if any pc hardware has this yet), there might be even more, microsoft's list is long but not fully clear about vld vs assisted.

[NikosD]

Excellent point.

I've done my homework and updated the first post for UVD 2.2. Added WMV and other info.
BTW, why MPC-HC doesnt' support WMV for UVD2.x ?

I think DXVA 2.0 only accepts MPEG2, VC-1, H.264, WMV codecs as DXVA valid.

UVD 2.2 does not support MPEG4 ASP (only VP4 does).
MPEG1 is too old.
I didn't know RMVB can be accelerated.

I'm almost ready to sue ATI about false advertising of MPEG2 bitstream support (does not exist) and H.264 L5.1 up to 4K x 2K support (does not exist)

Given the fact that UVD2.x was introduced in 2008, I think 2 years are more than enough to wait for ATI to fix MPEG2_VLD.

BTW, see my updated results for H.264 L5.1 1080p ReF 16.

With a driver tweak I got 6x performance!

There is so much "hidden" potential of UVD2 hardware that I really don't know what ATI is doing.

[renq]

What tweaks did You perform?

[tetsuo55]

There is a registry hack for mpeg2 dxva on ATI

[Underground78]

There is a registry hack for mpeg2 dxva on ATI

All ATI cards ?

[tetsuo55]

All ATI cards ?Probably only 4xx0 and newer

[mariner]

The .mov clip mentioned before is here:
http://rapidshare.com/files/416586142/1.Twinpeaks1080p30fpsRef2-27Mbps.mov.html

My results:

FFDShow DXVA: 47.17 fps
Microsoft DS: 47,75 fps
Microsoft MFT: 68,07 fps

DXVAChecker does not give Min/Avg/Max, has a problem with .mov container.
Just the final average number .

I haven't found any other sample.
If you have one, provide it.

This is a 60p VC1 WMV (http://www.megaupload.com/?d=2KAAU7RR) clip.

For testing MF decoders with 60p H264, you'll need to convert the panasonic clip into avi/mov/wmv.

[NikosD]

Probably only 4xx0 and newer

It's just a registry hack. It doesn't work.
I have already tried that.

[tetsuo55]

Ok, maybe they changed it with newer drivers? it worked for my hd4770

[NikosD]

It does show up in DXVA checker as ModelMPEG2_VLD but when you try to use it in a video player,it doesn't work.

@tetsuo

EDIT: As a good guy, I did my homework again. I enabled modelMPEG2_VLD and did the DXVA benchmarking.

It uses 40% of CPU and 98% !!! of GPU (GPU load).
This is not full bitstream support for MPEG2 my friend.
This is the definition of hardware assisted DXVA

[NikosD]

This is a 60p VC1 WMV (http://www.megaupload.com/?d=2KAAU7RR) clip.

For testing MF decoders with 60p H264, you'll need to convert the panasonic clip into avi/mov/wmv.

Very very interesting clip, althouth it took me 2 hours to download it.

If you find me another one with 1920x1080p60fps I will put in the first benchmark post.

This one is 1440x1088@60fps 40Mbps VC-1.Tough

No 60fps playback for my system with CPU only:

PotPlayer (40 fps to 60fps)
WMP12 (45 fps to 60 fps)
Splash Pro v1.2 (60fps but with sound glitches)

DXVA Artifacts with:

FFDShow
MPC-HC (Internal and standalone filters)
PotPlayer
VLC

Full 60fps playback without any artifacts and without any sound glitches only with DXVA by the following decoders:

ArcSoft codec
PowerDVD 10
Splash Pro v1.2

CPU usage ~8%
GPU usage ~0% with the above players

Benchmarks:

AMD MFT: 85/88/91
ArcSoft: 84/88/91
MPC-HC 84/87/91
FFDshow: 80/87/93

This is the first clip that can not be played by CPU and can be played by GPU for my system in a perfect way (60fps no sound glitches)

[tetsuo55]

I think "min" is the most interesting value, you dont want to drop any frames during video playback, its not like a game after all.

Also i think we should start adding cpu usage and gpu usage to each benchmark, so its clear when the cpu is being used

[mariush]

With Cyberlink 8 codecs in MPC-HC it says it plays MPEG 2 videos using DXVA, here's what DXVA Checker show:

http://savedonthe.net/image/673/mpeg2.png

The Leadtek one below though doesn't use DXVA (the option to enable is grayed out)

With the others, when videos play it shows about 6% gpu usage and about 5% cpu... but I think that 5% gpu usage is just resizing or something to do with the tv tuner running on the second display.

[NikosD]

I think "min" is the most interesting value, you dont want to drop any frames during video playback, its not like a game after all.

Well, I think in the way I chose to do benchmarking - because it unleashes the full potential of driver/hardware - we see speeds over the real (x1) playback, even if it's 60fps.
So in most cases min value of all codecs/decoders is well ahead of the valid framerate. (x1)

We really don't care if we have a difference in min values of 57 or 67 fps when the valid framerate is 24fps or 30fps.
There are no dropped frames. At all.

With one exception of course, H.264 1080p60Fps.
In that case none of the decoders is capable of min and avg values over valid framerate of 60fps. They are all below.


Also i think we should start adding cpu usage and gpu usage to each benchmark, so its clear when the cpu is being used

I have already done that in some of my posts, but only to show what the really good DXVA decoding is capable of.
When I say really good DXVA decoding I mean full bitstream decoding. And the really good or exceptionally good bitstream decoding is one with ~0% GPU usage and very low CPU usage < 10% for a typical Core2Duo.
Everything else is not a really good DXVA decoding (see Intel 4500MHD)
In all of my tests average GPU usage is ~0% (with spikes up to 7%) and CPU usage is ~7% (with spikes up to 13%)

[NikosD]

With Cyberlink 8 codecs in MPC-HC it says it plays MPEG 2 videos using DXVA, here's what DXVA Checker show:

The Leadtek one below though doesn't use DXVA (the option to enable is grayed out)

With the others, when videos play it shows about 6% gpu usage and about 5% cpu... but I think that 5% gpu usage is just resizing or something to do with the tv tuner running on the second display.

You seem to confuse DXVA versions (DXVA1, DXVA2) with DXVA implementations (assisted, full bitstream).
All the above implementations of MPEG2 DXVA, are assisted.

[NikosD]

@Tetsuo

How difficult/ possible is it for you to use AMD Fast (DnD) decoder for DXVA ?
It is the recommended decoder by ATI for all DXVA video formats (MPEG2,WMV,H.264,VC-1)
And the fastest one according to them and the benchmarks.

[NikosD]

@ namaiki

It seems that Intel H.264 DXVA decoding, misses one of the initial steps of the H.264 pipeline - a part of the frontend.
This part still uses CPU host.
To make it easier to understand, the situation is similar with VP2 and VC-1 DXVA. It is not completely accelerated by GPU.
So, I was completely right from the beginning, that something wasn't right about the definition of full support even though I have never used Intel IGP.
Wikipedia and Intel are lying about FULL bitstream DXVA H.264 support.
It is almost full. And almost full, is not full.

[Underground78]

What tweaks did You perform?

(filler)

[namaiki]

@ NikosD

It seems that Intel H.264 DXVA decoding, misses one of the initial steps of the H.264 pipeline - a part of the frontend.

Could you please link where you have read this? It might assist with this bug: https://sourceforge.net/apps/trac/mpc-hc/ticket/168

With a driver tweak I got 6x performance!
What driver tweak did you do? There are other users' posts that you have not responded to.

[NikosD]

Guys, I have done a lot of work to fill this thread with benchmark numbers.

I want some feedback, too.

I see no results for the samples provided at first post by VP2, VP3, VP4 and Intel ClearVideo users.

At least provide some CPU results (best of CoreAVC, DivX, FFMpeg-mt, DiAVC) for the 2nd clip of H.264.
The one with the extraordinary high GPU results.

[NikosD]

I found out a new 1920x1080p60 fps 40Mbps VC-1 clip and I have updated the first post with new numbers.

Something strange occured.
The numbers of the 1st VC-1 "easy" clip are almost the same with the 2nd VC-1 "hardest" clip.
No explanation.

[ranpha]

nVidia GT240 PureVideo4, result after three benchmark runs.

H.264 clips
----------
First clip:-

CoreAVC CUDA: 59/63/65
MPC-HC DXVA: 78/81/84
DivX 8 DXVA: 75/79/83
Microsoft DXVA DirectShow: 77/81/84

additionally:
Microsoft DXVA MediaFoundation: 78/106/183
nVidia Video Decoder MFT: 196/213/221


Second clip:

CoreAVC CUDA: 42/47/73
MPC-HC DXVA: 44/53/74
DivX 8 DXVA: 44/52/72
Microsoft DXVA DirectShow: 36/54/88


additionally: (remuzed into MP4)
Microsoft DXVA MediaFoundation: 42/63/72
nVidia Video Decoder MFT: 41/69/85


Third clip:

CoreAVC CUDA: 67/70/76
MPC-HC DXVA: 63/66/70
DivX 8 DXVA: 63/66/71
Microsoft DXVA DirectShow: 63/67/74

DXVA refused to enumerate the newly-remuxed-into-MP4 video. I don't know why, but the second clip works just fine after being remuxed.

VC-1
====

First clip:-
MPC-HC DirectShow (ModeVC1_VLD): 72/73/74
WMVideo Decoder DMO DirectShow (ModeVC1_IDCT - VMR9): 21/22/24
WMVideo Decoder DMO DirectShow (ModeVC1_IDCT - VMR7): 73/74/75
WMVideo Decoder DMO MediaFoundation (ModeVC1_IDCT): 61/71/77 (low CPU usage)
nVidia Video Decoder MFT: 73/74/77


Second clip:-
MPC-HC DirectShow (ModeVC1_VLD): 72/73/74
WMVideo Decoder DMO DirectShow (ModeVC1_IDCT): 19/22/25
WMVideo Decoder DMO DirectShow (ModeVC1_IDCT - VMR7): 68/73/77
WMVideo Decoder DMO MediaFoundation (ModeVC1_IDCT): 73/74/79 (high CPU usage)
nVidia Video Decoder MFT: 73/74/75

[renq]

A bit OT, but anyway:
How do you guys have/get MF codecs?
I have also Win 7, but haven't run into any of them...

[ranpha]

A bit OT, but anyway:
How do you guys have/get MF codecs?
I have also Win 7, but haven't run into any of them...

Windows ships with a few, most prominently the Microsoft H264 Video Decoder MFT. GPU vendors ships their own too. But without splitters they are almost useless really.

http://dl.dropbox.com/u/132004/mft_viewer.jpg

[mariner]

nVidia GT240 PureVideo4, result after three benchmark runs.

First clip:-

CoreAVC CUDA: 59/63/65
MPC-HC DXVA: 78/81/84
DivX 8 DXVA: 75/79/83
Microsoft DXVA DirectShow: 77/81/84

additionally:
Microsoft DXVA MediaFoundation: 78/106/183
nVidia Video Decoder MFT: 196/213/221


Second clip:

CoreAVC CUDA: 42/47/73
MPC-HC DXVA: 44/53/74
DivX 8 DXVA: 44/52/72
Microsoft DXVA DirectShow: 36/54/88


Third clip:

CoreAVC CUDA: 67/70/76
MPC-HC DXVA: 63/66/70
DivX 8 DXVA: 63/66/71
Microsoft DXVA DirectShow: 63/67/74

Greetings ranpha.

Can you also test the MF decoders on the third clip(you'll need to convert the H264 60p m2ts into mp4), as well as both the VC1 clips.

Many thanks and best regards.

[G_M_C]

[...]

With a driver tweak I got 6x performance!

There is so much "hidden" potential of UVD2 hardware that I really don't know what ATI is doing.

What tweaks did You perform?

(filler)

@ NikosD
[...]
What driver tweak did you do? There are other users' posts that you have not responded to.


Why is it that you dont want to answer the question NikosD ? I'm starting to get the feeling that it might be a hoax.

You realize that users will be reluctant to help you out in this thread when you refuse to answer some questions. Its all based on the principle of dialogue yo know. An seeing your general negativety i saw in the first and second page towards the guy with the Intel onboard, that simply did not understood right, makes me reluctant to even benchmark my system for you. Start communicating better, share light on you "secret" driver settings, and maybe this thread will fare better; i.e. you'll get more actual data.

[nevcairiel]

Windows ships with a few, most prominently the Microsoft H264 Video Decoder MFT. GPU vendors ships their own too. But without splitters they are almost useless really.

The NVIDIA decoder MFT is pretty awesome in speed.
Too bad we don't have any other components for Media Foundation. I'll be working on the splitter part, but that still leaves a nice and simple player to use - i don't even know if you could use the nvidia decoder in WMP, or if it forces its own decoder in the equation.

[NikosD]

Why is it that you dont want to answer the question NikosD ? I'm starting to get the feeling that it might be a hoax.

It's not a hoax. It's just a benchmark setting not a feature.
I have informed ATI and I'm waiting for Catalyst 10.9 to convert a benchmark setting to a useful setting.
The reason I put it on the benchmark table is just to show the potential of the hardware.

An seeing your general negativety i saw in the first and second page towards the guy with the Intel onboard, that simply did not understood right, makes me reluctant to even benchmark my system for you.

There is no general negativity by me. There is only special negativity.
My general attitude is positive and creative, that's why I created the thread and put a lot of effort in that.

Start communicating better, share light on you "secret" driver settings, and maybe this thread will fare better; i.e. you'll get more actual data.

Waiting for your results and less talking from you

[Underground78]

You still don't give us a way to try to reproduce your results ... Stop being so vague and maybe people will trust you and find an interest in what you are doing.

[NikosD]

nVidia GT240 PureVideo4, result after three benchmark runs.

First clip:-

additionally:
Microsoft DXVA MediaFoundation: 78/106/183
nVidia Video Decoder MFT: 196/213/221



Impressive.
It gives a push to the general idea that video player developers should try to use GPU vendor codecs than building their own.

[ranpha]

Greetings ranpha.

Can you also test the MF decoders on the third clip(you'll need to convert the H264 60p m2ts into mp4), as well as both the VC1 clips.

Many thanks and best regards.

Updated the original post.

[NikosD]

You still don't give us a way to try to reproduce your results ... Stop being so vague and maybe people will trust you and find an interest in what you are doing.

Don't try to change me. It's a take it or leave it.

[NikosD]

Updated the original post.

Ranpha, do you have any artifacts during normal playback of VC-1 clips in DXVA mode ?

[ranpha]

Ranpha, do you have any artifacts during normal playback of VC-1 clips in DXVA mode ?

No corruption with the twinpeaks clip, whatever DirectShow/MediaFoundation decoder I use.

The Devil May Cry clip has playback corruption problems with the MPC-HC decoder. No problem with the WMVideo Decoder DMO, and the MediaFoundation decoders.

[tetsuo55]

Hi NikosD,

The Scientific Method (http://en.wikipedia.org/wiki/Scientific_method) requires full disclosure of how the experiment was done, so that peers can confirm the results.
Since you made clear that you will not be providing full disclosure any time soon can you please add a description to your test result to make that clear? (e.g. stating that the "tweak" you used is probably unstable and you are in contact with AMD about it)

Most of us are used to ATI/AMD mostly ignoring customers/developers when it comes to DXVA/Video, so please forgive us if we have very low confidence in seeing any change in the drivers for years to come (some bugs reported 2+ years ago have stilll not been fixed).

[mariner]

nVidia GT240 PureVideo4, result after three benchmark runs.

H.264 clips
----------

Third clip:

CoreAVC CUDA: 67/70/76
MPC-HC DXVA: 63/66/70
DivX 8 DXVA: 63/66/71
Microsoft DXVA DirectShow: 63/67/74

DXVA refused to enumerate the newly-remuxed-into-MP4 video. I don't know why, but the second clip works just fine after being remuxed.




Greetings ranphs. Many thanks for updating test results.

Can you try this somewhat more forgiving Sanyo 60p clip (http://www.sanyo-dsc.com/products/lineup/dmx_hd2000/img/sample/movie_sample_hd2000_01.zip)?

Nikos, do you experience similar problem with you 5750?

[namaiki]

nVidia GT240 PureVideo4, result after three benchmark runs.

H.264 clips
----------
Third clip:

CoreAVC CUDA: 67/70/76
MPC-HC DXVA: 63/66/70
DivX 8 DXVA: 63/66/71
Microsoft DXVA DirectShow: 63/67/74


That one is interesting, as I would have expected CoreAVC("CUDA") to have lagged behind again.

[ranpha]

Greetings ranphs. Many thanks for updating test results.

Can you try this somewhat more forgiving Sanyo 60p clip (http://www.sanyo-dsc.com/products/lineup/dmx_hd2000/img/sample/movie_sample_hd2000_01.zip)?

Nikos, do you experience similar problem with you 5750?

MPC-HC DXVA - 89/90/90
CoreAVC CUDA - 71/72/73
DivX 8 DXVA - 87/90/91
Microsoft DXVA DirectShow - 90/90/91
Microsoft DXVA MediaFoundation - 92/114/209
nVidia Video Decoder MFT - 220/223/227

[Underground78]

It took me so long to download the samples from RapidShare that I've reupload all of them on MultiUpload to offer multiple hosting :

BBC_Planet_1080p24.mkv (96.05 MB) : http://www.multiupload.com/7KLO5Y976P
ken_ross_02-1920x1080_60fpsRef4.mts (68.87 MB) : http://www.multiupload.com/HVOPI5QCZJ
YouTube_-_Troya_HD_Test_4_4096_x_2304_23.97_fps.mp4 (13.98 MB) : http://www.multiupload.com/TFUGUHNM5E
1.Twinpeaks1080p30fpsRef2-27Mbps.mov (89.49 MB) : http://www.multiupload.com/2BQG03YA9U
Devil May Cry Gameplay.wmv (266.64 MB) : http://www.multiupload.com/HHH6SAM3NL
MP10_DigitalLife_10mbps.wmv (85.51 MB) : http://www.multiupload.com/KZQB5R9Z8S
twinpeaks.1080.wmv (45.81 MB) : http://www.multiupload.com/Y17WIJHRAS

[NikosD]

No corruption with the twinpeaks clip, whatever DirectShow/MediaFoundation decoder I use.

The Devil May Cry clip has playback corruption problems with the MPC-HC decoder. No problem with the WMVideo Decoder DMO, and the MediaFoundation decoders.

Microsoft does not support DXVA for VC-1 and WMV. I mean full bitstream DXVA support (VLD model).
It uses only CPU to play the VC-1, WMV files in WMP12 and some assisted DXVA if you force it for WMV in DXVAChecker (ATI).
That's why I haven't included Microsoft decoders for VC-1, WMV benchmarks.

[NikosD]

Greetings ranphs. Many thanks for updating test results.

Can you try this somewhat more forgiving Sanyo 60p clip (http://www.sanyo-dsc.com/products/lineup/dmx_hd2000/img/sample/movie_sample_hd2000_01.zip)?

Nikos, do you experience similar problem with you 5750?

I did an extended test with Media Foundation decoders (Microsoft, AMD) and found out that DXVAChecker has a lot of problems with AMD Fast (DnD) decoder.

The support of AMD MFT was included only at the last version 2.3.0.0 and it's a little buggy.

It crashes the application in a variety of files.

I had to unregister AMD MFT decoder in order to run the 1st H.264 clip benchmark with Microsoft MFT.

I have sent an email to the developer of DXVAChecker, hoping to find out a solution to his next version.

[ranpha]

Microsoft does not support DXVA for VC-1 and WMV. I mean full bitstream DXVA support (VLD model).
It uses only CPU to play the VC-1, WMV files in WMP12 and some assisted DXVA if you force it for WMV in DXVAChecker (ATI).
That's why I haven't included Microsoft decoders for VC-1, WMV benchmarks.

I've updated the original post to include ModeVC1_IDCT VMR7 results. For me it is just as good as ModeVC1_VLD, if not better. And that mode is usable in WMP12 (http://dl.dropbox.com/u/132004/dmc_wmp.jpg) and MPC-HC (http://dl.dropbox.com/u/132004/dmc_mpc.jpg) as well.

[G_M_C]

Don't try to change me. It's a take it or leave it.

I vote to leave it then.

[NikosD]

I have a new beta of DXVAchecker 2.4.0.0 and I'm doing some preliminary tests.
It seems that is using a different more accurate version of benchmarking that corrects some results to lower values.

I'll benchmark again all the clips, after the release of final 2.4.0.0 version, when publicy available.

[mariner]

Greetings Nikos.

1. You're right about DXVAChecker/AMD MFT being problematic. It would not even support interlaced VC1 content. Looking forward to your test results with version 2.4.0.0.

2. Thanks for the new 1920x1080@60p VC1 WMV. Great find. It would appear that all FFMpeg based decoders, including PotPlayer, have problem dealing with this.

Many thanks and best regards.

[mariner]

nVidia GT240 PureVideo4, result after three benchmark runs.

H.264 clips
----------

Third clip:

CoreAVC CUDA: 67/70/76
MPC-HC DXVA: 63/66/70
DivX 8 DXVA: 63/66/71
Microsoft DXVA DirectShow: 63/67/74

DXVA refused to enumerate the newly-remuxed-into-MP4 video. I don't know why, but the second clip works just fine after being remuxed.




Greetings ranphs. If you remux with just the video, did it work?

And thanks for testing the Sanyo clip.

[mariner]

Quick testing with VP3 engine. (9400IGP, W7x86)

1. DXVAChecker would not recognise nvdecoderMFT if latest 259.47 driver is installed. Had to revert to 195.62.

2. Both MS and Nvidia MFT were not listed when Panasonic 60p clip was selected for testing.

3. The Panasonic 60p clip played well at constant 60p using DS decoder, outperforming the UVD2. However, playback was choppy when using MPC with EVR/CP.

[ranpha]

Greetings ranphs. If you remux with just the video, did it work?

And thanks for testing the Sanyo clip.

I remux it into a video-only .mp4 file actually at first, and DXVA Checker won't enumerate it. The video plays fine with MPC-HC and WMP12 though, so I have no idea what went wrong.

[mariner]

Same problem here. Could be a DXVAChecker issue.

However, this Panasonic 60p clip edited with iMovie (zoo.mp4 (http://hdcam.web-pda.info/zoo%20-%2018512863.mp4)) will work . It also plays smoothly on a UVD2.

Thanks and best regards.

[mariush]

I've put all seven files here, so you can download them directly: http://mplayer.savedonthe.net/test_files/

OP if you wish, post the link in your first post, i'm not gonna delete these for a few weeks at least..

[NikosD]

Updated to 10.9 Catalyst.
Same results from first tests.

I'm going to wait for DXVAChecker 2.4 to update the test results of my first post.

I owe you the "secret" tweak.

Well, it's a registry hack.

You have to enable HWUVD_H264Level51Support and give it the value 1.


BTW, new DXVA profile for BluRay 3D - H264MVC with Catalyst 10.9

11572

[NikosD]

Update of the first post:

New AVCHD clip for benchmarking MFT decoders.
I have deleted the previous one.

Ranpha, please update your results for the new clip.

Underground78 and mariush, you may update download links for the new clip.

[mariush]

It's being transferred there as I write this. I may not access this thread every day or several times a day so if you wish, pm me whenever you add files.

[mariner]

Update of the first post:

New AVCHD clip for benchmarking MFT decoders.
I have deleted the previous one.

.....

I think you should keep the original Panasonic m2ts clip, and perhaps also upload a remuxed into mp4 version of the clip.

As I explained in my previous post, the zoo.mp4 has been edited with iMoive that makes it easier to decode. If you look at the media properties you'll find that it's quite different from the original clip.

[NikosD]

I think you should keep the original Panasonic m2ts clip, and perhaps also upload a remuxed into mp4 version of the clip.

As I explained in my previous post, the zoo.mp4 has been edited with iMoive that makes it easier to decode. If you look at the media properties you'll find that it's quite different from the original clip.

No, I put a whole different beast than zoo.
I put the birds clip.
We are talking about a 1920 x 1080p 60 fps 30Mbps clip.
It's a lot more difficult than zoo and more difficult than my first one.
Do your benchmarks and you will confirm this.
It's a special chosen clip like every clip I have in my first post

[NikosD]

There is a new DXVA DS codec in Catalyst 10.9(a)

It is called ATI MPEG Video decoder and supports DXVA for MPEG2, DX50, DIVX, H264, MP4V.

New video formats and some older ones.

Trying to use them and do some benchmarks

[Underground78]

You have to enable HWUVD_H264Level51Support and give it the value 1.

Where exactly should we put this key ?

Underground78 and mariush, you may update download links for the new clip.

Link : http://www.multiupload.com/PW2XUHHBJ6

[nevcairiel]

It is called ATI MPEG Video decoder and supports DXVA for MPEG2, DX50, DIVX, H264, MP4V.

Thats funny, because UVD2(.2) doesn't even support DX50, DIVX or MP4V (All MPEG-4 ASP variants). Only NVIDIAs VP4 supports those, and there is no DS decoder capable of working with it, afaik.

I guess it can try to do partial acceleration using Motion Compensation and iDCT using the shaders, but a comparison of those is always flawed because the host CPU does most of the hard work.

[Underground78]

It is called ATI MPEG Video decoder and supports DXVA for MPEG2, DX50, DIVX, H264, MP4V.

Where can I find them ? Are they available on XP ?

[NikosD]

Most of us are used to ATI/AMD mostly ignoring customers/developers when it comes to DXVA/Video, so please forgive us if we have very low confidence in seeing any change in the drivers for years to come (some bugs reported 2+ years ago have stilll not been fixed).

It seems that ATI did the work after all.
In Catalyst 10.9 there is a DXVA ATI MPEG video decoder (DirectShow) with H.264, VC-1?, WMV?, MPEG2, MPEG4 ASP video format support.

It may need further investigation, but looks promising.

You could take a look.

[tetsuo55]

Did they fix the bug with SD h264 playback though?

[NikosD]

No and probably never will.
You cannot have DXVA SD and HD Level 5.1 support at the same time.
They prefered to drop DXVA SD to support HD L5.1 DXVA.
It's better if you cannot have both.

The main DXVA problems for me are still:

1) UHD support
2) 60fps performance
3) MPEG2_VLD

Don't know if the new codec solves any of these.

[Underground78]

Did they fix the bug with SD h264 playback though?

What is this bug ? SD H264 works with DXVA here (HD2600XT) ...

I'm sorry to insist but where can I find this "ATI MPEG video decoder" ?

[NikosD]

DXVA for H.264 SD is still working with artifacts.

The codec is not hidden, you can find it in the usual way.

But I haven't found so far a way to use it in DXVA mode.

It's only 1 day old!

[nevcairiel]

You cannot have DXVA SD and HD Level 5.1 support at the same time.


NVIDIA can, so its not a strict DXVA limitation or anything.

[Underground78]

DXVA for H.264 SD is still working with artifacts.

I can't see any artifacts here. Are they obvious for you ?

The codec is not hidden, you can find it in the usual way.

Well, I can't. What is the file name of this filter ? Does it come with the display driver or the CCC ?

[NikosD]

NVIDIA can, so its not a strict DXVA limitation or anything.

Of course, I was only referring to ATI because it is a well known bug which has been transformed to a feature.

It's becoming more and more less important though, because of the low CPU usage of SD H.264.

[changturkey]

What video players are you guys using? I get artifacts with DXVA SD H.264 in VLC 1.1.4.

[Underground78]

What video players are you guys using? I get artifacts with DXVA SD H.264 in VLC 1.1.4.

MPC-HC lastest revision here (I can't test VLC since DXVA does not work under XP).

[littleD]

Regarding Mpeg-4 asp, both nvidia and ati claimed support for that format but as we know there is no dshow decoder with hardware acceleration for that. But i have read somewhere (dont remember exactly where) that AMD Fusion Media Explorer [beta] (http://sites.amd.com/us/athome/amdlive/fme/Pages/AMD-LIVE-Explorer.aspx) do that. Hint>>it can also work on Intel cpus after small hack. If anybody isn't feared of installing this software, then may test it. I cant install cause of some error.Regarding 4k support http://forums.amd.com/game/messageview.cfm?catid=260&threadid=136991&enterthread=y

[mariner]

No, I put a whole different beast than zoo.
I put the birds clip.
We are talking about a 1920 x 1080p 60 fps 30Mbps clip.
It's a lot more difficult than zoo and more difficult than my first one.
Do your benchmarks and you will confirm this.
It's a special chosen clip like every clip I have in my first post

The bird.mp4 clip has been transcoded to main@4.2 profile using HandBrake.

Can't get DZVAChecker to dsiplay any numbers when benchmarking DS decoders, so here are some screenshots of testing done on 9400IGP (VP3), W7x86 :

[NikosD]

But i have read somewhere (dont remember exactly where) that AMD Fusion Media Explorer [beta] (http://sites.amd.com/us/athome/amdlive/fme/Pages/AMD-LIVE-Explorer.aspx) do that. Hint>>it can also work on Intel cpus after small hack. If anybody isn't feared of installing this software, then may test it. I cant install cause of some error.

Asked for AMD CPU, I have C2D

Regarding 4k support http://forums.amd.com/game/messageview.cfm?catid=260&threadid=136991&enterthread=y

Same guy here:
http://www.rage3d.com/board/showthread.php?t=33965737&page=2

But I'm not interested in opinions of ATI beta testers.
I want someone to explain the official Catalyst 10.4 release notes regarding 4K x 2K support

[NikosD]

The bird.mp4 clip has been transcoded to main@4.2 profile using HandBrake.

Can't get DZVAChecker to dsiplay any numbers when benchmarking DS decoders, so here are some screenshots of testing done on 9400IGP (VP3), W7x86 :

Try different splitters. Cyberlink or Haali or Nero MP4 splitter
I had same problems solved by Haali splitter (I think it was Haali)

[changturkey]

MPC-HC lastest revision here (I can't test VLC since DXVA does not work under XP).

I just tested the X64 version of build 2561 of MPC-HC, artifacts still remain for SD H.264.

[littleD]

Asked for AMD CPU, I have C2D
Ok, you may try this (http://www15.zippyshare.com/v/18690889/file.html) and may need this (http://www.microsoft.com/downloads/en/details.aspx?FamilyID=dc614aee-7e1c-4881-9c32-3a6ce53384d9&displaylang=en#filelist).
Mhh.. and this software needs also windows indexing service enabled:/.

[NikosD]

Ok, you may try this (http://www15.zippyshare.com/v/18690889/file.html) and may need this (http://www.microsoft.com/downloads/en/details.aspx?FamilyID=dc614aee-7e1c-4881-9c32-3a6ce53384d9&displaylang=en#filelist).
Mhh.. and this software needs also windows indexing service enabled:/.

The patched version worked OK.
No DivX acceleration.
Too futuristic for me and a lot of garbage.
Sorry, but it's one of the worst programs I have ever used.
Uninstalled immediately.

[Blue_MiSfit]

Very impressive performance for the Microsoft Media Foundation decoder. Very impressive.

Derek

[NikosD]

Indeed.
But it really makes me wonder, so much work for both DirectShow and MediaFoundation DXVA H.264 decoders from Microsoft and so little support for their own VC-1, WMV ?

No bitstream DXVA support for VC-1, WMV3 in WMP12 in both DS and MFT.

It's a little weird that they don't support at the same level their own video format in WMP12.

[NikosD]

@Tetsuo
It seems that even under Catalyst 10.9a, ArcSoft decoder of the first post is capable of DXVA SD H.264 without artifacts.
But that decoder can't play L5.1 ReF 16 H.264 clips.
That's why it fails at the benchmark test of clip 2.

I think this is another proof that it's probably impossible to support DXVA SD and DXVA ReF 16 L5.1 clips at the same time in UVD2.x.

[tetsuo55]

At the very least its proof that ATI is making a mess of things.

Maybe we can leverage L5.1 in some way to produce correct results for SD

[NikosD]

After further investigation I can report for ATI codecs the following:

1) ATI Fast (DnD) decoder is a Media Foundation transcoder/ decoder.
It has both attributes of transcoder and DXVA decoder, but actually it works only as a transcoder. As a hardware decoder it's broken.
It's only installed under Win 7 and it is used by AVIVO transcoder in Catalyst.
If you try to use it in DXVAChecker as a decoder, it crashes the application with H.264 clips during enumeration.
VC-1 and WMV can be tested with AMD MFT decoder in DXVAChecker because they crash the application after the test.
The last non broken AMD MFT decoder was included in Catalyst 10.3 driver. This particular transcoder/decoder doesn't crash DXVAChecker but it is useless too as a DXVA decoder only. It crashes the application after the test.
Even if you try to make it usable in WMP12 replacing VC-1 and WMV WMDecoder MFT, it crashes WMP12 too.
So, forget AMD MFT decoder

2) The "new" ATI directshow MPEG video decoder, is not so new after all. It's in Catalys hotfix versions (like 10.7a or 10.9a). It seems that official Catalyst drivers don't include the decoder. You can find it in AVIVO optional download
It's the same decoder used in ATI cards back to 2006 and 2007!
It doesn't support DXVA MPEG4 ASP at all. Only CPU mode.
DXVAchecker cannot enumerate it in most video formats and when it does, the DXVA is not working.
So it seems a useless decoder like AMD MFT.

Conclusion:
ATI has no working DirectShow and MediaFoundation DXVA decoders.

[mariner]

nVidia GT240 PureVideo4, result after three benchmark runs.

H.264 clips
----------
Third clip:

CoreAVC CUDA: 67/70/76
MPC-HC DXVA: 63/66/70
DivX 8 DXVA: 63/66/71
Microsoft DXVA DirectShow: 63/67/74

DXVA refused to enumerate the newly-remuxed-into-MP4 video. I don't know why, but the second clip works just fine after being remuxed.


Greetings ranpha.

Just like to confirm with you that MPC can render the Panasonic 1080/60p m2ts clip smoothly in EVR/CP using DXVA.

Many thanks and best regards.

[NikosD]

@Mariner(VP3)
@Ranpha(VP4)
@Namaiki(VP2, Intel)

Hello.
Is it hard for you to benchmark all the clips of the first post as it is right now, with your video engines ?
You can put your results in a post and I can put your tests to the first post in order to have them all in one place.

Thanks

[NikosD]

Updated first post results.
Vista x86 tests with CPU and GPU usage.
DXVA1 and DXVA2 versions added.
MPEG2 assisted results.

CPU average ~3%, GPU average ~0% for every bitstream format supported (H.264, VC-1, WMV3)
MPEG2 GPU average ~0% (No GPU shaders used - Only UVD and CPU for DXVA MPEG2)

[NikosD]

@Tetsuo
I know that VC-1 and WMV3 are both members of WMV9 and WMV3 is fully compliant with VC-1.
How much work is it for a developer to make the MPC - Video Decoder capable of using the VC-1 decoder for WMV3, too?
I have no idea.

[tetsuo55]

interesting, i think directshow does treat them differently, do they work with dxva in vlc? or any of the comercial codecs?

[NikosD]

Yes, VLC decodes VC-1 and WMV3 in the exact same way.
The same is true for ArcSoft and Cyberlink in some versions.
It would be useful to have MPC-HC as an all-in-one filter for DXVA.

[NikosD]

New kid on the block - UVD 3.0

Here is a screenshot of DXVA capabilities
11686

New codecs:

1) H.264MVC_VLD

2) MPEG4ASP_VLD

3) MPEG2_VLD

The first one (BluRay 3D) works in HD 5000 series too, but without HDMI 1.4a - like in HD 6000 series - is practically useless for HD 5000.

The second (DivX, Xvid etc), although had VP4 support months ago, it is the first time that it is actually supported. The support of Nvidia was paper support.
The new DivX beta codec has ATI DXVA support.

The third has a long story...

I don't have a 68xx card in my hands, but I would really like to know H.264 performance/capabilities - UHD , 1080p60, SD

[NikosD]

Catalyst 10.10 brings:

1) Removal of H264_MVC for HD 5000 introduced in Catalyst 10.9 (But it works in PowerDVD 10 Ultra Mark II as it worked before the Catalyst 10.9)

2) WMV HD support in Win 7 (It was there from the beginning - I see no difference, another trick from ATI)

3) Improved Dynamic contrast algorithm

4) Support for Blu-ray 3D - HD 6800 series

5) Stereo 3D gaming for HD 5000, 6800 series

I don't think I'm going to redo the tests for this driver.

[tetsuo55]

They seem to have changed a lot in the driver this time, results might be different when compared to older drivers.
Retesting would be welcome

[mesostinky]

I've found this to be a very interesting thread. I guess most people aren't complaining because they have fast enough cpu's and don't even worry about dxva anymore? Anyway I just ordered a low powered amd system, a dell zino w/P340/onboard vid, and after reading here about lacking DXVA implementations and seeing DXVA benchmarks like this (http://www.bit-tech.net/hardware/graphics/2010/05/24/amd-880g-graphics-performance-review/7) I wonder if I'd made a mistake? I originally was going to get a atom/ion system but just ordered a Zino instead. So sorry for taking the thread slightly off course but does anyone have an opinion if AMD should be avoided for low end HTPC systems due to issues talked about in this thread? Thanks

[NikosD]

Well, it seems that ATI/AMD finally did it.

In Catalyst 10.12 they fixed the H.264 DXVA SD bug.

You can play your favorite anime in SD with full DXVA support in both DS and MFT decoders.

Also, the WMV HD (WMV3) and VC-1 support in Win 7 introduced in Catalyst 10.10, simply meant they added a new Media Foundation decoder with an obvious name: AMD Playback Decoder MFT capable for full DXVA decoding of WMV3 and WVC1 files.

This new MFT decoder is used automatically by WMP12 only and accelerates successfully only progressive VC-1 (not interlaced - many artifacts)

The performance of that MFT decoder is the same as of AMD Fast (DnD) MFT decoder, but without crashing DXVAchecker.

Maybe the most interesting thing I found out in Catalyst 10.12 (perhaps it was there in earlier versions too) is the HUGE performance increase (x4) in Microsoft H264 Video Decoder MFT.

The difficult clip of birds (clip 3) -1080p60 fps gave me a max of 460fps!, an average of 224 fps and a min of 64.

Check out the first page to realize that people in ATI have taken their job of improving DXVA performance seriously in the latest Catalysts.

[tetsuo55]

wow that is good news, too bad you do not have a 6xxx card to add divx results with

[hwti]

The difficult clip of birds (clip 3) -1080p60 fps gave me a max of 460fps!, an average of 224 fps and a min of 64.


It seems very strange :
- benchmark results are very bad with DS (Microsoft DTV-DVD decoder) : Avg 24 / Min 23 / Max 30
- both DS and MFT do not manage to play correctly (55fps average)

Perhaps the decoder skips frames in benchmark mode, making the result wrong.

[NikosD]

It seems very strange :
- benchmark results are very bad with DS (Microsoft DTV-DVD decoder) : Avg 24 / Min 23 / Max 30
- both DS and MFT do not manage to play correctly (55fps average)

Perhaps the decoder skips frames in benchmark mode, making the result wrong.

Try WMP12 and the built-in FPS counter of WMP12.
Give me your results.

[hwti]

Try WMP12 and the built-in FPS counter of WMP12.
Give me your results.

Most of the video is played at 43fps (it starts at 40-50fps, there is a big drop to <20fps after 3s, then it goes up and stabilises).

[NikosD]

Card?
OS?
Catalyst version?

[hwti]

HD5870
Win7 64bit
Catalyst 10.12 (8.801-101125a-109686E)

[NikosD]

Try a clean system.

I can't think of anything else

[hwti]

I checked the number of DXVA2_DecodeDeviceBeginFrame events in DXVAChecker Trace Log.
The video has 2487 frames.

WMP12 : 2322 events
So it skips a few frames, which is not surprising since it can't play at full speed.

MF benchmark : 2146 events (fps results : avg 312 min 61 max 973)
341 frames have been skipped.
The maximum fps is random, between 350 and 1000 fps, it could perhaps come from skipped frames being counted as decoded.

DS benchmark : 2487 events
So all frames are decoded


The 24fps average framerate with DS benchmark comes from the fact it wants to play at 25fps ("(@25fps)" text in blue).
It seems to come from Haali Media Splitter.
In MPC-HC, looking in the "Pin Info" tab of the decoder (ffdshow with ffmpeg-mt):
- with Haali Media Splitter, the framerate is not present on input, and is 25fps on output (but it still plays at the correct framerate)
- with internal MP4 source, the framerate is 59.94fps on both pins

So I tried the DS benchmark with the standalone "MPC - MP4 Splitter".
It works, but the fps and time stay at 0 during the benchmark (even if I enable the preview, which shows the video is correctly decoded).
But in "play" mode, the fps and time are displayed !

[NikosD]

Interesting research.

I suggest to forward your results to DXVA Checker developer. I know he was in the process of making a new version. Your results of MFT benchmarking mode would be useful.

I have a beta of DXVA checker 2.4.0.0 and the DS benchmark mode gives 52/56/60 and the MFT benchmark mode gives similar results with 2.3.0.0

In my last email to him, I was complaining about the "strange" results of built-in WMP12 fps counter he is using as reference.

I suggested to him to use FRAPS.

I suggest to you also to use FRAPS while playing the clip in WMP12 and come back with your results.

[hwti]

In WMP12 FRAPS gives 57-60fps, so some decoded frames are displayed several times, or WMP12 counter is wrong.

[NikosD]

New MFT DXVA codec for UVD3 in Catalyst 11.6

It's called AMD Hardware MFT Playback decoder and accelerates in hardware MPEG-4 ASP:

MP4S, M4S2, MP4V, DIVX, DX50, XVID

[TheElix]

Wow! Will it be possible to use it in MPC-HC for those formats?

[nevcairiel]

Wow! Will it be possible to use it in MPC-HC for those formats?

No, MPC-HC does not support MFTs, and most likely never will. Only real player that can use them is WMP.

BTW, why are there two of this thread?

[NikosD]

This is the first one created for DXVA in general.

The other is for DXVA H.264 only vs CPU.