CoreCodec/H.264 Codec "CoreAVC"

[squid_80]

That would be right if I were using VMR9, but it happens with EVR and EVR Custom. When playing TFF content it's clear that the renderer (or another filter located after CoreAVC like ffdshow) is receiving BFF flags. When you force TFF then all is right. Again this has nothing to do with VMR. Already discussed and confirmed with tests a few pages back.
EVR contains most of the same bugs as VMR9.

I'd like to know how you're examining the flags. Are you aware they can change at any time during playback, even be set differently for each frame? CoreAVC specifies the stream as BFF while connections are being made but passes the correct field order downstream as soon as it starts decoding.

[Inspector.Gadget]

Using CoreAVC with the version of nvcuvid.dll included in the newest DGAVCIndexNV generates a brief period of image corruption on a 720p video that doesn't occur when CUDA acceleration is disabled. This does not appear on seeking, but only on initially opening the video and beginning playback and depending on the file appears as a flash of pure bright green for a few frames or blocking/distortion problems lasting several seconds. I'm using an 8600M GT with Windows Vista Home Premium 32-bit. The details of the file in question are below; I'm using Haali's Matroska Splitter with MPC-HC as a frontend, ffdshow as an audio decoder, and the VMR9 renderless renderer. Edit: This problem also appears with Haali's renderer, so it seems to be renderer-independent.

General
Complete name : C:\File.mkv
Format : Matroska
File size : 4.38 GiB
Duration : 1h 42mn
Overall bit rate : 6 091 Kbps
Encoded date : UTC 2009-02-08 09:55:59
Writing application : mkvmerge v2.4.1 ('Use Me') built on Dec 5 2008 18:30:05
Writing library : libebml v0.7.7 + libmatroska v0.8.1

Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4.1
Format settings, CABAC : Yes
Format settings, ReFrames : 5 frames
Muxing mode : Container profile=Unknown@4.1
Codec ID : V_MPEG4/ISO/AVC
Duration : 1h 42mn
Bit rate : 4 304 Kbps
Nominal bit rate : 4 607 Kbps
Width : 1 280 pixels
Height : 544 pixels
Display aspect ratio : 2.35
Frame rate : 23.976 fps
Resolution : 24 bits
Colorimetry : 4:2:0
Scan type : Progressive
Bits/(Pixel*Frame) : 0.258
Writing library : x264 core 66 r1099 c0be810
Encoding settings : cabac=1 / ref=5 / deblock=1:0:0 / analyse=0x3:0x133 / me=umh / subme=7 / psy_rd=1.0:0.0 /
mixed_ref=1 / me_range=16 / chroma_me=1 / trellis=1 / 8x8dct=1 / cqm=0 / deadzone=21,11 / chroma_qp_offset=-2 / threads=12 / nr=0 / decimate=1 / mbaff=0 / bframes=3 /
b_pyramid=1 / b_adapt=1 / b_bias=0 / direct=1 / wpredb=1 / keyint=250 / keyint_min=25 / scenecut=40(pre) / rc=2pass /
bitrate=4607 / ratetol=1.0 / qcomp=0.60 / qpmin=10 / qpmax=51 / qpstep=4 / cplxblur=20.0 / qblur=0.5 / ip_ratio=1.40 / pb_ratio=1.30 / aq=1:1.00

Audio
ID : 2
Format : DTS
Format/Info : Digital Theater Systems
Codec ID : A_DTS
Duration : 1h 42mn
Bit rate mode : Constant
Bit rate : 1 536 Kbps
Channel(s) : 6 channels
Channel positions : Front: L C R, Surround: L R, LFE
Sampling rate : 48.0 KHz
Resolution : 24 bits
Language : English

Text
ID : 3
Format : UTF-8
Codec ID : S_TEXT/UTF8
Codec ID/Info : UTF-8 Plain Text
Language : English

Is there another version of nvcuvid I should try? I've hunted through the installation files for the latest appropriate driver and not found one. :thanks:

[TheShadowRunner]

Here CUDA works, Geforce 8500 / XP / 182.05, however shows artifacts not present with software decoding.
CPU use is greatly reduce though, going from 60% to around 15% when CUDA is enabled.
The artifacts:
http://videoff7.free.fr/cuda.jpg
http://videoff7.free.fr/soft.jpg
Later,

TSR

[deekey777]

I definitely know that 8800GTS has limited DXVA support, but CUDA is different stuff, not related to DxVA and according to NVIDIA (http://www.nvidia.com/object/cuda_learn_products.html) is supported on my board.
So I believe the problem is elsewhere...

http://developer.download.nvidia.com/presentations/2008/NVISION/NVISION08_ImageVideoCUDA_web.pdf
Go to the page 29/30.

[squid_80]

List of supported cards as listed in the CoreAVC Installer:
NVIDIA GeForce GTX 260/280, 9800, 9600, 9500, 8800 GT, 8700, 8600, 8500, 8400, Tesla S1070/C1060, Quadro FX 3700, Quadro FX 3600M, Quadro FX 1700/FX 570/ NVS 320M/FX 1600M/FX 570M/FX 370/NVS 290/NVS 140M/NVS 135M/FX 360M/NVS 130M and higher.
Unfortunately the 8800 Ultra, GTX and GTS miss out.

[BetaBoy]

squid... I added that list to the first post in this thread, the CC forum thread, and the KB on the support center.

[samepaul]

http://developer.download.nvidia.com/presentations/2008/NVISION/NVISION08_ImageVideoCUDA_web.pdf
Go to the page 29/30.

Ok, so CoreaAvc 1.9 doesn't really have decoding "CUDA-code", but rather tries to use VP via CUDA API instead of DxVA API? That's all?

[BetaBoy]

TheShadowRunner the artifacts/blockiness is known thx for the feedback.

I also want to that everyone for the amazing amount of details and results so far... it definitely is helping track down the issues.

[nm]

Ok, so CoreaAvc 1.9 doesn't really have decoding "CUDA-code", but rather tries to use VP via CUDA API instead of DxVA API? That's all?
Yep, that's all.

[STaRGaZeR]

EVR contains most of the same bugs as VMR9.

I'd like to know how you're examining the flags. Are you aware they can change at any time during playback, even be set differently for each frame? CoreAVC specifies the stream as BFF while connections are being made but passes the correct field order downstream as soon as it starts decoding.

http://forum.doom9.org/showthread.php?t=104277&page=205

Since you were there you know what I'm talking about. ffdshow fixed that problem with all renderers, it works perfectly now with all H.264 sources. The flags can be examined with ffdshow's internal deinterlacers. They're not affected by any renderer, and clearly shows that CoreAVC flags TFF content as BFF. You can output YV12 from CoreAVC and put ffdshow as a postprocessing filter handling the deinterlacing process. If any deinterlacer is set to Auto, it'll behave the same as the "faulty" renderers, which indicates it's not the renderer's fault. If this is not CoreAVC sending the wrong flags downstream I don't know what it is. Force TFF and everything is OK. Use Auto (it'll use the field order that is passed to the deinterlacer, in this case CoreAVC's field order) or Force BFF and you'll get the typical back and forth motion. This is with samples that have all their frames or interlaced macroblocks (if it's MBAFF) flagged as TFF. On a side note PAFF samples with TFF flags in all their interlaced frames behave the same.

[Jay Bee]

VMR9 is broken. It even changes field order randomly when you seek.

Not true since XP SP3.

[vucloutr]

List of supported cards as listed in the CoreAVC Installer:

Unfortunately the 8800 Ultra, GTX and GTS miss out.

Funny thing is that the 8800 GTS 512 is also supported (G92 chip).
People usually refer to these models as 8800 GTS with 512MB or 1024MB RAM.
I guess one can blame Nvidia for their confusing naming scheme.

[ACrowley]

Mh... i dont have a Nvidia Card but a ATI HD4870

Something is strange with 1.9.0.0. BBC-HD 1080i MBAFF Files are stuttering/shaking with Hardware deinterlacing. It works fine with 1.8.5.0

Ah...the Stuttering is caused by Haali Splitter + CoreAVC
Works perfect with Gabest/MPC Mpeg Splitter and also Haali+ MPC Videodecoder (DXVA+HW Deinterlacing active)

Im not sure if it was working with older Haali Version.

[Cyber-Mav]

initial tests show that there is no decoding speed difference between a 8600gt and a 8800gt i have. both yield in similar levels of cpu usage, but thats to be expected since both of the cards use the same VP2 processor.

back to testing...

[Mixer73]

TheShadowRunner the artifacts/blockiness is known thx for the feedback.

So BB is this just a factor of hardware decoding of the streams or is it something that can improve?

[BetaBoy]

Mixer73.... its something we are working on fixing in our CoreAVC CUDA support and not a problem with CUDA... well at least that's what it looks like atm. So far since the release we have fixed the reported 'threading issue' and the 'blockiness after seeking' bugs.

[Dark Eiri]

BetaBoy, could you have a look into that issue with 10 ref-frames videos? They were encoded with the old HQ-slowest profile on MeGUI. Curiously, videos encoded with the HQ-Insane profile, 16 ref-frames, are perfectly decoded. And also, videos encoded with old revisions of x264 fall back to software decoding.

[Kurtnoise]

Hey,

I'm looking for some benchmarks between CoreAVC Cuda enabled vs MPC Dec DXVA. Could be great to have some comparisons...

[BetaBoy]

Hey,

I'm looking for some benchmarks between CoreAVC Cuda enabled vs MPC Dec DXVA. Could be great to have some comparisons...

I agree ;-)

[ranpha]

Video
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L5.1
Format settings, CABAC : Yes
Format settings, ReFrames : 8
Codec ID : V_MPEG4/ISO/AVC
Duration : 1mn 22s
Bit rate : 9613 Kbps
Nominal bit rate : 10000 Kbps
Width : 1920 pixels
Height : 1088 pixels
Display aspect ratio : 16/9
Frame rate : 59.940 fps
Colorimetry : 4:2:0
Scan type : Progressive
Bits/(Pixel*Frame) : 0.080
Writing library : x264 - core 56 svn-667C



CoreAVC CUDA
User: 25s, kernel: 2s, total: 27s, real: 66s, fps: 177.3, dfps: 73.9
User: 24s, kernel: 2s, total: 27s, real: 64s, fps: 181.5, dfps: 76.5
User: 24s, kernel: 2s, total: 27s, real: 66s, fps: 178.4, dfps: 74.5

CoreAVC 1.9 software mode
User: 14s, kernel: 0s, total: 14s, real: 41s, fps: 342.2, dfps: 118.7
User: 13s, kernel: 0s, total: 13s, real: 41s, fps: 355.3, dfps: 119.6
User: 15s, kernel: 0s, total: 15s, real: 41s, fps: 319.8, dfps: 119.2

MPC-HC 908 DXVA
User: 220s, kernel: 0s, total: 220s, real: 221s, fps: 22.4, dfps: 22.4
User: 215s, kernel: 0s, total: 216s, real: 216s, fps: 22.9, dfps: 22.8
User: 215s, kernel: 0s, total: 215s, real: 216s, fps: 22.9, dfps: 22.8

MPC-HC 908 software mode
User: 28s, kernel: 2s, total: 30s, real: 205s, fps: 160.3, dfps: 24.1
User: 27s, kernel: 3s, total: 30s, real: 208s, fps: 160.1, dfps: 23.7
User: 26s, kernel: 3s, total: 29s, real: 206s, fps: 166.4, dfps: 24.0


Video
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4.1
Format settings, CABAC : Yes
Format settings, ReFrames : 4
Codec ID : V_MPEG4/ISO/AVC
Duration : 1mn 31s
Nominal bit rate : 1200 Kbps
Width : 1280 pixels
Height : 720 pixels
Display aspect ratio : 16/9
Frame rate : 23.976 fps
Colorimetry : 4:2:0
Scan type : Progressive
Bits/(Pixel*Frame) : 0.054
Writing library : x264 - core 59 r859M ce13bb6


CoreAVC CUDA
User: 5s, kernel: 1s, total: 6s, real: 16s, fps: 315.5, dfps: 135.2
User: 5s, kernel: 1s, total: 6s, real: 16s, fps: 338.2, dfps: 130.9
User: 6s, kernel: 1s, total: 7s, real: 16s, fps: 295.0, dfps: 133.2

CoreAVC 1.9 software mode
User: 3s, kernel: 0s, total: 3s, real: 6s, fps: 698.1, dfps: 346.5
User: 3s, kernel: 0s, total: 3s, real: 6s, fps: 638.1, dfps: 358.0
User: 3s, kernel: 0s, total: 3s, real: 6s, fps: 715.9, dfps: 351.7

MPC-HC 908 DXVA
User: 30s, kernel: 0s, total: 30s, real: 30s, fps: 72.3, dfps: 71.6
User: 28s, kernel: 0s, total: 28s, real: 28s, fps: 78.4, dfps: 78.1
User: 28s, kernel: 0s, total: 28s, real: 28s, fps: 77.4, dfps: 76.9

MPC-HC 908 software mode
User: 5s, kernel: 0s, total: 6s, real: 32s, fps: 357.0, dfps: 67.2
User: 5s, kernel: 0s, total: 5s, real: 33s, fps: 385.3, dfps: 66.1
User: 5s, kernel: 0s, total: 5s, real: 32s, fps: 378.1, dfps: 68.0


Video
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L5.1
Format settings, CABAC : Yes
Format settings, ReFrames : 6
Codec ID : V_MPEG4/ISO/AVC
Duration : 23mn 40s
Nominal bit rate : 1041 Kbps
Width : 720 pixels
Height : 480 pixels
Display aspect ratio : 1.850
Frame rate : 29.970 fps
Standard : NTSC
Colorimetry : 4:2:0
Scan type : Progressive
Bits/(Pixel*Frame) : 0.101
Writing library : x264 - core 56 svn-667C


CoreAVC CUDA
User: 63s, kernel: 16s, total: 79s, real: 133s, fps: 534.2, dfps: 319.3
User: 62s, kernel: 16s, total: 79s, real: 132s, fps: 537.9, dfps: 322.2
User: 57s, kernel: 14s, total: 72s, real: 130s, fps: 585.5, dfps: 325.5

CoreAVC 1.9 software mode
User: 21s, kernel: 0s, total: 21s, real: 52s, fps: 1968.2, dfps: 807.5
User: 20s, kernel: 0s, total: 20s, real: 52s, fps: 2047.9, dfps: 810.7
User: 21s, kernel: 0s, total: 21s, real: 51s, fps: 1947.2, dfps: 822.1

MPC-HC 908 DXVA
User: 174s, kernel: 0s, total: 175s, real: 176s, fps: 242.7, dfps: 241.5
User: 175s, kernel: 0s, total: 176s, real: 176s, fps: 241.9, dfps: 241.4
User: 173s, kernel: 0s, total: 174s, real: 174s, fps: 244.8, dfps: 244.2

MPC-HC 908 software mode
User: 53s, kernel: 3s, total: 57s, real: 253s, fps: 743.2, dfps: 168.1
User: 55s, kernel: 3s, total: 59s, real: 258s, fps: 718.4, dfps: 165.1
User: 54s, kernel: 3s, total: 57s, real: 242s, fps: 735.2, dfps: 175.9

nVidia 9800GT with 182.05 beta drivers under Windows 7 beta.
CPU is Phenom X4 9500 2.2Ghz and RAM is 4GB DDR2 667.

All numbers show that the CUDA decoder is faster. Now if only the decoder can play those files without artifacts that isn't seen in CoreAVC software mode or MPC-HC in DXVA and software mode, that would have been better.

MPC-HC software mode seems to be slower than DXVA mode because it only uses 1 core (even if I set it to 4 in its property page), unlike CoreAVC that can use all 4 cores.

[Kurtnoise]

could you add also values for CoreAVC w/o Cuda & MPC Dec w/o DXVA ?

:thanks:

[Gleb Egorych]

Some bug with interlaced content support. I tried BBC Galapagos Blu-ray, MediaInfo says it's MBAFF. The main video is progressive while titles in the end are interlaced.

Cyberlink H.264 decoder with DXVA enabled and "auto-select" deinterlacer plays the video at 50fps and bobs video (with bob artifacts).
CoreAVC 1.8.5 and 1.9.0 with "hardware" deinterlacer play it at 50fps, but discard "the second field" of the titles. Looks like them weave the main video and double the first field of the titles.

With bob deinterlacer CoreAVC plays both fields, the result seems to be indentical to Cyberlink DXVA. But bob deinterlacer affects picture quality (line-doubling).

[ajp_anton]

MPC-HC software mode seems to be slower than DXVA mode because it only uses 1 core (even if I set it to 4 in its property page), unlike CoreAVC that can use all 4 cores.At least when looking at the number of threads in task manager, it goes up when raising the number of threads in MPC-HC (all the way up to 6). Plus with two threads it can play things that it can't with one.

[lexor]

Am I reading that wrong, or is CoreAVC software faster than CUDA in those tables ranpha posted? We are supposed to look at dfps, right?

[ajp_anton]

Am I reading that wrong, or is CoreAVC software faster than CUDA in those tables ranpha posted? We are supposed to look at dfps, right?Yes, with a quad core, and with 100% CPU usage instead of 10%.

[Malow]

"prefer cuda acceleration" is greyed on my pc.

i have a Asus M3N-HT deluxe, chipset nforce 780a, nvidia 8xxx compilant onboard video.

all other softwares detect and use hardware decoding perfectly.

[BetaBoy]

If its greyed out then either its not a compliant NVIDIA CUDA card (not all 8xxx cards are compliant.. for example the 8800GTX is not compliant) or you have not installed the 182.05 beta drivers. When in doubt look at the supported CUDA card list:

NVIDIA GeForce GTX 260/280/290/295, 9800, 9600, 9500, 8800 GT, 8700, 8600, 8500, 8400, Tesla S1070/C1060, Quadro FX 3700, Quadro FX 3600M, Quadro FX 1700/FX 570/ NVS 320M/FX 1600M/FX 570M/FX 370/NVS 290/NVS 140M/NVS 135M/FX 360M/NVS 130M and higher.

[ajp_anton]

When in doubt look at the supported CUDA card list:8800GTS 512 (and 1024 if those exist?) MB is still missing.

[Malow]

If its greyed out then either its not a compliant NVIDIA CUDA...
i reeeealy belive it is, cause it have a 8400GS core, all others hardware-based decoders can do gpu decoding (cpu usage 1%) and it even works with hardware encoding softwares for CUDA GPUs, like BadaBOOM Media Converter.

i know, some 8xxx does not are cuda-enabled, (and also does not have full purevideo-HD capabilities) but mine has.

H.264 Decode Acceleration
H.264 Decode Acceleration with IDCT and CAVLC/CABAC
VC-1/WMV Decode Acceleration
VC-1/WMV Decode Acceleration with IDCT
MPEG-2 Decode Acceleration
High-Quality Scaling
MPEG-2 Spatial-Temporal De-Interlacing
MPEG-2 Inverse Telecine

question: how coreavc detect compatible GPUs?

[imk]

Will there ever be support for the G80 8800 series cards?

[squid_80]

i reeeealy belive it is, cause it have a 8400GS core, all others hardware-based decoders can do gpu decoding (cpu usage 1%) and it even works with hardware encoding softwares for CUDA GPUs, like BadaBOOM Media Converter.Have you installed the 182.05 driver?

[Malow]

ahhhhh, was the driver.... i was using 185.20, as it is "higher" than 182.05, i was thinking it was ok. but 185.20 is in fact older than 182.05...

now back to the decoder, with cuda enabled i got 15~20% of cpu usage on 720p clip (athlon x2 5200+). its normal?

[chros]

You don't need the beta driver, you can use the latest WHQL: v181.22 ... At least, it works for me ...
(System: GF 9600GT, WinXP SP3)

[squid_80]

You don't need the beta driver, you can use the latest WHQL: v181.22 ... At least, it works for me ...
(System: GF 9600GT, WinXP SP3)
The 181.22 driver by itself does not contain all the needed components. Perhaps you've installed one of the DG***IndexNV programs.

[leeperry]

If its greyed out then either its not a compliant NVIDIA CUDA card (not all 8xxx cards are compliant.. for example the 8800GTX is not compliant) or you have not installed the 182.05 beta drivers. When in doubt look at the supported CUDA card list:
you can add the 9600GSO/8800GS...as I have one of these ;)

[hajj_3]

are those of you who are saying your card isnt supported copying the .dll to /windows/system32 ? as the box will be greyed out until you do so. I now have the option enabled after doing that. I've got an Nvidia 9400GT. You can add that card to the supported list as in all of the Geforce 9 Series have CUDA support: http://www.nvidia.com/object/product_geforce_9400gt_us.html

[Sharc]

Works here with 9600GS / 182.05.
However, few green frames when starting playback and ugly blocks for about 1 second when seeking or changing audio track.

[squid_80]

are those of you who are saying your card isnt supported copying the .dll to /windows/system32 ?
This is not the recommended procedure, it is preferred to install the 182.05 driver or higher instead. This takes care of installing nvcuvid.dll (and ensures it is an up-to-date build) and also avoids confusion between system32/syswow64 on 64-bit systems.

[hajj_3]

true but the 182.05 drivers are beta, i like alot of other nvidia users get 4bit colour and 800x600 res when upgrading nvidia drivers so have to revert to previous drivers so i try not to update nvidia drivers too often and certainly not to betas. Wish nvidia fixed their updating drivers problem but that may never happen.

[vucloutr]

I made this list of graphics devices that should be supported from "nv_disp.inf" of the 182.05 beta driver.

supported graphics cards (G84, G86, G92, G94, G96, G98 Core):
GeForce 8300 GS, 8400, 8400 SE, 8400 GS, 8500 GT, 8600 GS, 8600 GT, 8600 GTS, 8800 GS, 8800 GT, 8800 GTS 512
GeForce 9300 SE, 9300 GE, 9300 GS, 9300M GS, 9400 GT, 9500 GS, 9500 GT, 9600 GS, 9600 GSO, 9600 GSO 512, 9600 GT, 9800 GT, 9800 GTX, 9800 GTX+, 9800 GX2
GeForce GTX 260, GeForce GTX 280, GeForce GTX 285, GeForce GTX 295
Quadro NVS 290, NVS 420, NVS 450,
Quadro CX
Quadro FX 370, FX 370 LP, FX 570
Quadro FX 1700, FX 3700, FX 4700 X2, FX 4800, FX 5800
Quadro VX 200
Quadroplex 2200 S4
Tesla C1060

supported integrated chipsets:
GeForce 8200, 8300
GeForce 9200, 9300, 9400
Quadro FX 470


definitely _not_ supported (G80 Core):
GeForce 8800 GTX, 8800 GTS, 8800 Ultra
Quadro FX 5600, FX 4600
Tesla C870

[BetaBoy]

8800GTS 512 (and 1024 if those exist?) MB is still missing.

ajp_anton... thx for the report. When we get confirmation that it works and there are not different generations of the cards then we will add it/them.

[hajj_3]

betaboy, can we expect a 1.9.1 version with all the bugs reported fixed next week?

[Cyber-Mav]

8800gts 512 is a 128 stream processor version of the 8800gt and both cards use the G92 core, afterall the 8800gts 512 is a 9800gtx.

[Malow]

I made this list of graphics devices that should be supported from "nv_disp.inf" of the 182.05 beta driver.
a few onboard video chipset should work too (like mine)

nForce 780a SLI
nForce 750a SLI

nForce 790i SLI
nForce 780i SLI
nForce 750i SLI

how about this list?
http://www.nvidia.com/object/cuda_learn_products.html

You don't need the beta driver, you can use the latest WHQL: v181.22 ... At least, it works for me ...
(System: GF 9600GT, WinXP SP3)
didn't work for me :(

[nm]

how about this list?
http://www.nvidia.com/object/cuda_learn_products.html
That includes G80 cards that don't have VP2.

[Malow]

8800gts 512 is a 128 stream processor version of the 8800gt and both cards use the G92 core, afterall the 8800gts 512 is a 9800gtx.
there is something different with GeForce 8800 GTX and GeForce 8800 GTS...

http://i39.tinypic.com/hurog4.png

That includes G80 cards that don't have VP2.
well, i guess we are looking for a specific capability of cuda GPUs... being cuda is not enough :)

[~bT~]

http://en.wikipedia.org/wiki/CUDA#Supported_GPUs

http://en.wikipedia.org/wiki/PureVideo#Table_of_PureVideo_.28HD.29_GPUs

[88keyz]

Here is another sample of the the video blocking present on the CUDA accelerated CoreAVC vs. the DXVA accelerated MPCDecoder which does not show the blocking effect.

CoreAVC 1.9 with CUDA enabled
http://img17.imageshack.us/img17/8783/coreavcv19withcudaxi4.th.png (http://img17.imageshack.us/img17/8783/coreavcv19withcudaxi4.png)

MPCDecoder with DXVA enabled
http://img294.imageshack.us/img294/7/mpcdecoderwithdxvaom4.th.png (http://img294.imageshack.us/img294/7/mpcdecoderwithdxvaom4.png)

Looks like MPCDecoder with DXVA is still the way to go for hardware accelerated H.264 video files.

[Malow]

http://en.wikipedia.org/wiki/CUDA#Supported_GPUs

http://en.wikipedia.org/wiki/PureVideo#Table_of_PureVideo_.28HD.29_GPUs
nice...

my onboard video shoud be G98/VP3 i guess, cuz it can play 1080p VC1 in MPC-HC/DXVA using ... 0.00% cpu... im happy with my onboard video ; :)

[BetaBoy]

betaboy, can we expect a 1.9.1 version with all the bugs reported fixed next week?

No timetables but as I've said we expected a few 1.9.x releases... we are working hard on issues reported and have already fixed a number of them.