View Full Version : DGAVCDecNV 1.0.13: GPU decoding on Nvidia
Guest
4th January 2009, 16:09
I've noticed two small errors in DGAVCIndex (and NV) User Manuals Thank you for pointing these out.
Guest
4th January 2009, 16:10
The d2v files included the colorimetry info. Is this information also included in the dga files? It's shown in the info dialog during the project creation, but I cannot find it in the dga file... It's not in the DGA file but it is in the log file.
yesgrey
4th January 2009, 17:27
It's not in the DGA file but it is in the log file.
Thanks. I think it would be safer to also put it in the dga, in case we delete the log... but ok, I just have to be carefull and only delete the log after all the work has been done.:)
hajj_3
4th January 2009, 18:39
has development started on an ATI version of this app yet neuron or are you waiting for their equivalent of CUDA to improve or waiting for API's that will allow you to create an ATI version?
This app is looking very nice indeed.
Guest
4th January 2009, 19:04
has development started on an ATI version of this app yet neuron or are you waiting for their equivalent of CUDA to improve or waiting for API's that will allow you to create an ATI version? How do you expect me to do anything in the absence of an API?
hajj_3
4th January 2009, 19:17
ahh, didnt realise they hadn't released the API's, are they planning to release them or have they not released any info about their plans with API's?
Guest
4th January 2009, 19:22
Off topic for this thread and you'd have to ask them. I am not privy to their business plan.
Chefkoch_ico
6th January 2009, 09:24
regarding license.
I just donated and got the license.
I am going to return my new graphics card and get a different one ("Point of View" vs EVGA, both 9600GT) since its too loud and the fan not controlable. Not automatic or by software.
Do I need a new license for it afterwards?
Best regards
Guest
6th January 2009, 14:49
You won't need a new license.
BTW, my BFG 9600GT was making noise too, but it was a subtle "chirping". It turned out to be a small piezoelectric speaker intended to alarm an undervoltage condition. I'm sure there was no undervoltage. Anyway, I just covered the little exit hole of the speaker and it's inaudible now.
Guest
9th January 2009, 21:39
1. Added support for the new audio types found on BluRay disks.
2. Improved the audio delay calculation.
3. Removed the Treat I as IDR option as it is now always enabled.
4. Machine ID now not tied to system partition, so machine ID changes will be rare.
5. Updated the CUVID server to 1.0.3.
http://neuron2.net/dgavcdecnv/dgavcdecnv.html
madshi
10th January 2009, 08:23
3. Removed the Treat I as IDR option as it is now always enabled.
Just for my own interest: Doesn't this result in artifacts during seeking? I mean, I've been told that if you seek to an I (non IDR) frame with h264, you're in danger of getting artifacts. Aren't we supposed to look for "recovery point SEIs"?
Guest
10th January 2009, 08:56
If the GOP is open, then the decoder backs off by one GOP and then decodes enough frames to get back to the seeked-to frame. This is exactly the same thing done in DGMPGDec.
rica
11th January 2009, 03:24
Neuron,
you've given me a lot of homework to do at the weekend :)
Guest
11th January 2009, 03:29
rica, you've helped me so much in the past with your testing and feedback, I want to acknowledge that. :thanks:
rica
11th January 2009, 03:43
You are welcome my friend.
I always appreciated your endless efforts; to be grateful is on my side.
Thanks a lot one more time btw.
I gonna try to do my best.
Denner
12th January 2009, 11:53
Hi.
Stumbled upon this thread in my search for quicker x264 encoding, and this might help me do that, am I correct ?
If i understand correctly DGAVCDecNV is a filter ( or is it a program ) that will frame serve for me leaving all of my Intel quad core's processing power free to do the x264 encoding, and in that way help me achieve faster encoding speeds ?
I got a Geforce 9400 GT card, but is that a fast enough graphics card or wont it help me at all ?
I will gladly donate ( pay ) for the filter, but I am a little concerned that I have misunderstood what DGAVCDecNV is and how to use it :confused:
Guest
12th January 2009, 14:50
It's discussed several times in this thread. Bottom line: speedup will be quite modest on fast systems because decoding requires a small fraction of the total decoding+encoding load. On some slower systems it can give a useful speedup. The motivation for using the GPU is to get correct decoding, access to the PureVideo postprocessor, and support for VC1.
Denner
12th January 2009, 16:21
Ok I am sorry if it was a stupid question, just wanted to ask before I buy it from you :p
rica
12th January 2009, 20:41
Denner, i can suggest you to use DGAVC so as to have an idea.
It is a free, software based decoder+frame Server using Libav Codec instead of nVidia's HW based decoder.
I'm about to open a topic on BD retranscoding via DGAVCNV and DGVC1NV.
Denner
12th January 2009, 22:43
Cool and thanks, I will try it out :)
Guest
16th January 2009, 05:51
It's early days and all that, but I thought you might be interested to know that I just decoded and displayed my first MKV file with AVC video in DGAVCIndexNV. I adapted low-level MKV parsing code provided to me by Haali. :thanks:
There's still a lot of work to do, especially with DGAVCDecodeNV and audio demuxing, but it looks like there's nothing to stop me from implementing full MKV support. With DivX backing MKV+AVC now, I think it will be a useful addition to the DG tools.
In case you're wondering, I made my own MKV file with mkvtoolnix. :)
Sharc
16th January 2009, 07:28
.... but it looks like there's nothing to stop me from implementing full MKV support. With DivX backing MKV+AVC now, I think it will be a useful addition to the DG tools.
In case you're wondering, I made my own MKV file with mkvtoolnix. :)
Cool, thanks!
ultratoto14
16th January 2009, 18:30
It's early days and all that, but I thought you might be interested to know that I just decoded and displayed my first MKV file with AVC video in DGAVCIndexNV. I adapted low-level MKV parsing code provided to me by Haali. :thanks:
There's still a lot of work to do, especially with DGAVCDecodeNV and audio demuxing, but it looks like there's nothing to stop me from implementing full MKV support. With DivX backing MKV+AVC now, I think it will be a useful addition to the DG tools.
In case you're wondering, I made my own MKV file with mkvtoolnix. :)
Congratulations.
rica
17th January 2009, 20:58
It's early days and all that, but I thought you might be interested to know that I just decoded and displayed my first MKV file with AVC video in DGAVCIndexNV.
Congrats :thanks:
Clumpco
19th January 2009, 08:02
Donald, please credit me with a little more intelligence than that.....
..... All this makes capturing/encoding a very linear process and reduces the benefit of the high framerates obtained.
I cannot capture from a sat card while indexing/encoding.
I cannot index another file while encoding or capturing.
I cannot edit a .ts while indexing/encoding.
I have come back to my original post to apologise for being rather shirty. I was struggling with glitches, sync problems and generally being a bear of little brain at the time. I'm sorry.
The server/decoder is excellent software and I would be lost without it.
Since I posted I have changed my sat card (got a Skystar HD2 - highly recommendable) and this new one will work in the bottom PCI slot. The previous one would only work in the 2nd slot which put it right next to my double-width 9600GT. This was leading to overheating problems which explains why I could not index/frameserve while capturing - obviously the extra heat from the 9600GT engine during this operation was just enough to cause the sat card to glitch seriously.
Keep up the good work!
xopowo
19th January 2009, 09:22
I have come back to my original post to apologise for being rather shirty. I was struggling with glitches, sync problems and generally being a bear of little brain at the time. I'm sorry.
The server/decoder is excellent software and I would be lost without it.
Since I posted I have changed my sat card (got a Skystar HD2 - highly recommendable) and this new one will work in the bottom PCI slot. The previous one would only work in the 2nd slot which put it right next to my double-width 9600GT. This was leading to overheating problems which explains why I could not index/frameserve while capturing - obviously the extra heat from the 9600GT engine during this operation was just enough to cause the sat card to glitch seriously.
Keep up the good work!
Don't sweat it man. I've been using this since day one of its release and I can't live without it either. There is one thing in particular that bugs the crap out of me that happens with CUVIDServer, but I just deal with it since there is no other option. So, I know where you're coming from. I guess the next time when you get frustrated, make sure that what you "can" and "cannot" do is troubleshot at your level before you blame the software.
The issue I have with CUVIDServer is 99.999% soimething with it though, or so I would think :)
popper
19th January 2009, 12:12
ahh, didnt realise they hadn't released the API's, are they planning to release them or have they not released any info about their plans with API's?
its OT here (you can open up and directly link a new thread related to ATI UVD ASIC use if you like)but for a one time reference here, i asked Bridgeman about this UVD ASIC subject and pointed him to neuron2's and BetaBoys VS2 ASIC API work....heres the basic information known at this time...
"Mr bridgeman the ATI/AMD Executive in charge of the current code/documentation releases and ATI market push, has been asked to work with the open code initatives and provide the equivanent NV/VS2 ATI UVD API documentation were he stated in the very professional developer relations manner last week
"bridgeman:we are going to look into opening up UVD, I just can't make any commitments until we have actually gone through the investigation and it won't be quick. ...."
and he confirmed ATT after i asked will it take months, "bridgeman:For open source, yes,"
the one point of light at the moment (but probably not relevant in this case unless the open/closed code developers of NV cuda/VS2 apps have a personal interest in also producing UVD code in the future for [non] windows platform), is this, Bridgeman also said "I expect fglrx will have it sooner" when refering to that long month on months wait for the open source vendors.
potentially that means any org or business looking to use and/or port the closed ATI/AMD fglrx driver or use any of the "UVD" capabilitys could potentialy get hold of the required API documentation, work with Bridgeman, and impliment it into their products...."
CruNcher
19th January 2009, 12:50
Popper it's typical for the Industry it's slow (@ releasing stuff they have developed the last 2 years and finished laying around) and fooling the consumer when ever it can, but Nvidia is currently far far ahead of ATI they even run them over in Video which is was ATIs dominating Market space for years they run over ATIs Xilieon platform also with Tegra @ it's back and the Ion Platform too and Ati has Paper laying around and talks about "we investigate" that's pathetic just pathetic it shows that this company is not able currently to react fast enough to the market and it's competitors, i wonder how his explanation impacted their share on wall street :D
Guest
19th January 2009, 14:12
There is one thing in particular that bugs the crap out of me that happens with CUVIDServer, but I just deal with it since there is no other option. What is it about the server that bugs you?
rack04
19th January 2009, 14:23
Only "problem" that I've noticed with CUVIDServer is that if I choose to cancel an encode CUVIDServer hangs. I have to ctrl-alt-del to get it to close.
Esurnir
19th January 2009, 22:12
Only "problem" that I've noticed with CUVIDServer is that if I choose to cancel an encode CUVIDServer hangs. I have to ctrl-alt-del to get it to close.
I confirm, when you stop x264 (by a -hard stop- like two time control C and not let it end "gently") cuvidserver hang and no other encode can be done. Happen in my case too.
menlvd
23rd January 2009, 23:21
neuron2
any chance to get worked in xp x64 CUVIDServer?
Guest
23rd January 2009, 23:54
I don't have XP 64.
Audionut
24th January 2009, 10:35
Do you have a changelog from 1.0.8 to 1.0.10?
edit: cuvidserver works fine in vista and windows 7 64bit. Or are you requesting a 64bit build menlvd?
edit: nevermind, it's in the release notes. Thanks Don.
menlvd
24th January 2009, 14:19
Do you have a changelog from 1.0.8 to 1.0.10?
edit: cuvidserver works fine in vista and windows 7 64bit. Or are you requesting a 64bit build menlvd?
edit: nevermind, it's in the release notes. Thanks Don.
no, in my wxp x64 sp2 even don't start :(
jeffy
24th January 2009, 14:44
@menlvd: Do you have a supported GPU?
menlvd
24th January 2009, 14:50
@menlvd: Do you have a supported GPU?
eepp, dgavcindexnv working fine, using 181.22 forceware drv
Guest
24th January 2009, 17:57
Good news...MKV support for video is working all the way through to frame accurate serving via DGAVCDecodeNV.
I want to do some regression testing for ES and TS and then I'll make a test release.
Then I need to add audio support.
blubberbirne
24th January 2009, 19:29
good news neuron2
deets
25th January 2009, 10:55
this is the only bit of kit i found that can handle itvhd streams, thanks very much :)
Guest
25th January 2009, 16:56
Here's a test version for MKV support. No audio yet. Feedback will be appreciated. Use the same CUVID server and nvcuvid.dll as for 1.0.10. This version uses a revised DGA file format, so you'll have to remake your existing projects if you want to open them with this version.
http://neuron2.net/dgavcdecnv/dgavcdecnv1011test.zip
When reporting problems, please provide a stream sample that I can use to duplicate the issue.
Deinorius
25th January 2009, 17:11
Am I right, that you are testing it with NVversion and after then adopting it to the libav-version?
Guest
25th January 2009, 17:26
I haven't decided yet. I don't know if I want to do anything more with the libavcodec version. I'd rather spend my time making a version that works with CoreAVC.
Deinorius
25th January 2009, 17:34
If libavcodec or CoreAVC (I'd love it :) ) I don't care. But like much more others I don't own a CUDA featuring nvidia card. This is the critical part of the story.
woah!
25th January 2009, 20:33
mkv processing seems ok even though it seems to choose 29.97fps unless forced to 23.976fps which the mkv is (original avc video from a bluray + 1 audio stream). also dont skip through the mkv with the slider, it will hang. just load the mkv and process it straight away. you cant skip through the finished avs script either.
rebkell
25th January 2009, 20:50
I haven't decided yet. I don't know if I want to do anything more with the libavcodec version. I'd rather spend my time making a version that works with CoreAVC.
That would be nice, I'm happy with my ATI cards, and I would definitely be all over a CoreAVC app.
Guest
25th January 2009, 23:17
mkv processing seems ok even though it seems to choose 29.97fps unless forced to 23.976fps which the mkv is (original avc video from a bluray + 1 audio stream). also dont skip through the mkv with the slider, it will hang. just load the mkv and process it straight away. you cant skip through the finished avs script either. How can I fix it if you don't give me a stream?
I even asked explicitly for that in my post!
Guest
25th January 2009, 23:27
MKV test version is withdrawn for further testing.
rica
25th January 2009, 23:37
neuron,
I've been busy with reinstalling everything today so i haven't found the opportunity giving it a go.
I was about to be ready to try.
Guest
26th January 2009, 00:14
Thanks. I found it's broken for GOPs smaller than 16 frames. In the shop for repairs...
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.