VideoHelp Forum




+ Reply to Thread
Page 4 of 4
FirstFirst ... 2 3 4
Results 91 to 95 of 95
  1. I noticed in his command line --deinterlace true => so this is done by cuda.exe in this case ?
    yes

    Sorry for the offtopic, but is their development kit closed for those other operations like scaling etc, ?
    the encoder library only offers a few things, see: http://docs.nvidia.com/cuda/samples/3_Imaging/cudaEncode/doc/nvcuvenc.pdf
    looking at it, one could probably use some internal resizer, but since there is no option to select which method it uses, you can bet on it that it's nearest neighbour (so not useful at all, which is why I never bothered to add it to the wrapper).

    How do other programs leverage cuda cores for those other operations ?
    they use cuvid (the VPX chipset) instead of the nvcuenc encoder chip.
    DGDecNV is basically a wrapper around cuvid with additional code which acts as a splitter to open the container (the additional code was based on DGMPGDec and later was extended to handle additional container)
    The CUDA SDK also includes an example application to playback stuff, so in theory it shouldn't be to hard to combine Cuda.exe and the decoder application from the sdk into one,...
    In praxis, I simply lack the time and motivation to look into it.
    Especially since DGDecNV does a good job; I even asked neuron2 a while back if DGDecNV could be used as a decoder, see: http://forum.doom9.org/showthread.php?p=1406924.
    -> if someone with some C/C++ knowledge and some motivation wants to have a go at it I'll happily share the source code of the cuda.exe wrapper with him, as long as he'll share his results for free with me and the community.

    Cu Selur
    Quote Quote  
  2. Thanks for the clarification Selur .

    ...We return you now back to your regular scheduled programming...
    Quote Quote  
  3. Member
    Join Date
    Nov 2002
    Location
    United States
    Search Comp PM
    Originally Posted by jagabo View Post
    Originally Posted by DarrellS View Post
    I think I get around 148fps at superfast preset.
    I happen to have a Q6600 too. I just ran the same encoding test I ran earlier on the i5 2500K (DVD MPEG 2). I got 172 fps at superfast. 127 fps at veryfast. That was a simple AviSynth script:

    Code:
     Mpeg2Source("VTS_01_1.d2v")
    with the following x264 command line:

    Code:
     x264.exe --preset=veryfast --crf=18 --keyint=100 --sar=10:11 --output %1.mkv %1

    I ran the 640x360 mpg again using my (264 MKV Superfast no audio encoder set)...

    --crf 18 --preset superfast --tune film --demuxer raw --input-csp i420 --input-res %(width)x%(height) --fps %(fpsnum)/%(fpsden) -o "%(tempvideofile)" -

    ...and got 360.69 fps. What was the last cuda encode, 347fps? Hhhmmmmm.

    EDIT: Ultrafast is over 400fps. The log file doesn't give me the encode time because it happens so fast (18 seconds) but looking at the encoding window shows 417fps. Both the SF and UF encodes leaves 4 to 6 rows of blurred pixels like some old FLV files I worked on a few years back but the ultrafast encode got rid of the macroblocking that was in the original mpg.
    Last edited by DarrellS; 4th Apr 2013 at 10:16.
    Quote Quote  
  4. Member
    Join Date
    Nov 2002
    Location
    United States
    Search Comp PM
    The pixels issue on the x264 encodes turned out to be MPC-HC. I unchecked all internal filter and the problem went away.
    Quote Quote  
  5. With CRF encoding bitrates really shoot up with superfast and ultrafast. When I had a slower computer I usually stuck with veryfast. These days I don't do a lot of encoding, have a faster computer, and I mostly work with standard definition sources, so I usually use slow. I don't care if it takes 30 minute to convert a movie vs. 10 minutes.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!