Recherche avancée

Sur d’autres sites (10274)

• swscale/aarch64 : use multiply accumulate and increase vector factor to 4

  17 novembre 2019, par Sebastian Pop

  swscale/aarch64 : use multiply accumulate and increase vector factor to 4
  This patch implements ff_hscale_8_to_15_neon with NEON fused multiply accumulate
  
  and bumps the vectorization factor from 2 to 4.
  
  The speedup is of 25% on Graviton1 A1 instances based on A-72 cpus :
  $ ffmpeg -nostats -f lavfi -i testsrc2=4k:d=2 -vf bench=start,scale=1024x1024,bench=stop -f null -
  
  before : t:0.040303 avg:0.040287 max:0.040371 min:0.039214
  
  after :  t:0.032168 avg:0.032215 max:0.033081 min:0.032146
  The speedup is of 39% on Graviton2 m6g instances based on Neoverse-N1 cpus :
  
  $ ffmpeg -nostats -f lavfi -i testsrc2=4k:d=2 -vf bench=start,scale=1024x1024,bench=stop -f null -
  
  before : t:0.019446 avg:0.019423 max:0.019493 min:0.019181
  
  after :  t:0.014015 avg:0.014096 max:0.015018 min:0.013971
  Tested with `make check` on aarch64-linux.
  Signed-off-by : Sebastian Pop <spop@amazon.com>
  
  Reviewed-by : Jean-Baptiste Kempf <jb@videolan.org>
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libswscale/aarch64/hscale.S

• FFmpeg transcode GIF into Mp4 and Mp4 to AVI using GPU

  9 octobre 2023, par Cristian

  I'm trying to convert GIF animated to mp4 and mp4 to AVI with FFmpeg.

  


  I started to use just the CPU, but I have to process millions of GIFs/mp4 content pieces. So, I started to have a lot of errors processing them, and it ended as a bottleneck. Therefore, I'm trying to use GPU to process the videos.

  


  Converting GIF to mp4 with CPU, I run the following command :

  


  ffmpeg -i animated.gif -movflags faststart -pix_fmt yuv420p -vf "scale=trunc(iw/2)*2:trunc(ih/2)*2" video.mp4


  


  Using the GPU I'm trying the following :

  


  ffmpeg
  -y
  -hwaccel nvdec
  -hwaccel_output_format cuda
  -i gifInputPath
  -threads 1
  -filter_threads 1
  -c:v h264_nvenc
  -vf hwupload_cuda,scale_cuda=-2:320:240:format=yuv420p
  -gpu 0
   mp4VideoPath


  


  The above command generates an exit status 1.

  


  The following is the dmesg command log

  


  Converting mp4 videos to AVI videos I'm running the following command

  


  ffmpeg
-i videoInputPath
-vcodec rawvideo
-pix_fmt yuv420p
-acodec pcm_s16le
-ar 44100
-ac 2
-s 320x240
-r 4
-f avi
aviOutputVideoPath


  


  For GPU I tried :

  


  ffmpeg
 -y
 -hwaccel cuda
 -hwaccel_output_format cuda
 -i videoInputPath
 -threads 1
 -filter_threads 1
 -c:a pcm_s16le
 -ac 2
 -ar 44100
 -c:v h264_nvenc
 -vf hwupload_cudascale_cuda=-2:320:240:format=yuv420p
 -r 4
 -f avi
 -gpu 0
 aviOutputVideoPath


  


  The following is the dmseg output is log

  


  


  1. What should be the best command for converting the GIF into Mp4 and Mp4 into AVI based on CPU configuration using the GPU(Amazon Nvidia t4) for best performance, low CPU, and moderated GPU consumption ?

     


  


  2. What are the best suggestions to Process these content pieces concurrently using GPU ?

     


  


  


  Note : I'm using Golang to execute the FFmpeg commands.

  


• ffmpeg QSV hardware encoder with x11grab screen capture

  11 janvier 2020, par Toby Eggitt

  I believe I have built ffmpeg with support for my motherboard’s Intel graphics processor chip, but I have not succeeded in showing this working in any way. My goal is to use it for screen capture (the ffmpeg I built does capture screen successfully using the software encoding, but this is far too slow to be useful—it manages about 12fps at a very modest quality).

  My main problem—I think—is that I don’t know how to use these encoders, the examples I found all fail, which makes me suspect that what I’ve built is broken in some way. However, I also have no idea how I can verify that I built this correctly, but the following are true :

  • The five components that I built to get to this all compiled without
    errors (they were libva, gmmlib, intel-media-driver, libmfx, and
    ffmpeg
  • The output of ffmpeg -encoders includes four encoders with _qsv in
    their names including h264_qsv
  • Most of the commands I have tried result in output of this form :

      [h264_qsv @ 0x55ef1dc72040] Low power mode is unsupported
  
      [h264_qsv @ 0x55ef1dc72040] Current frame rate is unsupported
  
      [h264_qsv @ 0x55ef1dc72040] Current picture structure is unsupported
  
      [h264_qsv @ 0x55ef1dc72040] Current resolution is unsupported
  
      [h264_qsv @ 0x55ef1dc72040] Current pixel format is unsupported
  
      [h264_qsv @ 0x55ef1dc72040] some encoding parameters are not supported by the QSV runtime. Please double check the input parameters.
  
      Error initializing output stream 0:0 -- Error while opening encoder for output stream #0:0 - maybe incorrect parameters such as bit_rate, rate, width or height

  I have the impression this thing might be fussy about many parameters of this sort but have no idea where to find out what it would like. Any suggestions at all, how to verify it, or better yet, how to issue a command that captures screen and encodes with the hardware, would be most welcome.