audio glitch on src=480p60fpsVP9 media playback on Samsung_galaxy_S5 |
|||
Issue descriptionVersion: Android, samsung OS: 52.0.2723.0 on Samsung-gaglaxy-S5 device, there is audio glitch in playing src=480p60fpsVP9 video. It not repro on other video (e.g 720p60fps, 360p60fps) thougk, it NOT repro on mse=480p60fps video either. recorded video is at https://pantheon.corp.google.com/storage/browser/avlab/videostack/2016-05-10-22:40:28:370817 Dale, I assigned this bug to you. Let me know if you need this device to investigate. I'll check this device from lab give to Frank. What steps-will reproduce the problem? (1) launch 52.0.2723.0 on Samsung-galaxy-S5, (2) navigate to http://172.17.32.252/videostack/src/video.html?file=480p60fpsVP9_titanfall_sync.webm (3) listen to audio part What is the expected output? smooth audio playback What do you see instead? glitch on a few places
,
May 16 2016
Hmm, I'd guess 60fps material is starving the audio renderer even at 480p. We only spin up an extra thread for >= 720p decoding. That said, I'm surprised that 480p decodes are enough to saturate the media thread on the S5 though. +fgalligan, +jzern in case there's something odd about our encodes / if that's less than expected performance. Frank, can you collect a trace from this device so we can see how long each video decode is taking?
,
May 17 2016
So you want us to collect a trace in clank on an S5? Or try a raw decode? Then we will need to know which S5? But I don't understand why you think that is a perf issue. It works fine on 720p60, even with 4 threads for 720p vs 2 threads for 480p, 720p has more than 2x the pixels of 480p, so in general if it can play the 720p60 then it should be able to play the 480p60. But even if we assume it has something to do with that 480p file, it works fine with MSE. Are they using 2 different decoders at this point?
,
May 17 2016
Err, sorry fgalligan@, I should have said liberato@ who is also a Frank :)
,
Jun 24 2016
I seems not repro this on 6/23 A/V automation run.
,
Jul 13 2016
|
|||
►
Sign in to add a comment |
|||
Comment 1 by yini...@chromium.org
, May 13 2016