Announcement

Collapse
No announcement yet.

Faster VP9 Multi-Threaded Video Decoding Patch Lands In Libvpx

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Faster VP9 Multi-Threaded Video Decoding Patch Lands In Libvpx

    Phoronix: Faster VP9 Multi-Threaded Video Decoding Patch Lands In Libvpx

    Performance optimizations are always great presents to see in open-source projects around the holidays (well, any time of the year for that matter). Libvpx today picked up another optimization for helping out with VP9 video decoding...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Nice, but does anyone use libvpx for decoding? ffmpeg's decoder is generally much faster and has supported multihreading since... forever?

    Comment


    • #3
      nice, i hope you compare the decode performance to ffmpeg too.

      Comment


      • #4
        I would find a comparison between ffvp9 and dav1d to be quite meaningful.

        Comment


        • #5
          When faced with something "new" in VP9, he manages to dig up the corresponding feature/commit in AV1. Just wow! I call that good journalism!

          Few journalists come close to Michael's technical insight and ability.
          Last edited by andreano; 21 December 2018, 08:03 PM.

          Comment


          • #6
            Originally posted by andreano View Post
            When faced with something "new" in VP9, he manages to dig up the corresponding feature/commit in AV1. Just wow! I call that good journalism!

            Few journalists come close to Michael's technical insight and ability.
            Are you okay?

            The commit message doesn't elaborate on any estimated performance gains out of this tile superblock row based multi-threading. However, when the tile superblock row based multi-threading was added to the AV1 encoder earlier this year, for reference, it yielded a speed-up of about 2.9x for four threads.
            He's telling you the kinds of performance gains you can expect to see.

            Comment

            Working...
            X