User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:100.0) Gecko/20100101 Firefox/100.0

Steps to reproduce:

Only after setting media.hardware-video-decoding.force-enabled=true does Firefox use hardware decoding for AV1 videos. I tested this out on YouTube.

Actual results:

After setting the above to true, Firefox uses the hardware decoder. Everything up to 4K 60FPS playback is fine, but 8K 60FPS playback is choppy even though stats_for_nerds shows no dropped frames.

Expected results:

I should not have had to tweak media.hardware-video-decoding.force-enabled as per my understanding.

Zaggy1024, can you reproduce this?

I've ran mozregression as suggested, added as an attachment.

It's kinda weird because 88a481adf83070d1cdcc2bb074f6545b5dfe97e7 from 2022-03-19 23:09:55.398000 is ok, but 88a481adf83070d1cdcc2bb074f6545b5dfe97e7 from 2022-03-19 23:07:24.391000 (earlier date) is not, and then again 64ec4c736bb4b643e413173ca7403cca4d269671 from 2022-03-19 23:10:49.212000 is not ok (latest).

Hopefully this is helpful.

Ok, Tomasz, I think your bug seems unrelated to the AV1 issue. Can you file a new one?

Odd, I can't seem to reproduce this on the beta branch, AV1 decoding works correctly for me on my Intel Xe laptop. Are you able to run some logs for me? If you run Firefox from the command line with args like this:

firefox -MOZ_LOG="MediaSource:5,MediaFormatReader:5,MediaDemuxer:5,PlatformDecoderModule:5,MediaDecoder:5,sync,timestamp" -MOZ_LOG_FILE="%userprofile%\Documents\firefox_log"

you should have some files in your documents folder, firefox_log.child-1.moz_log should be the GPU process, which would hold most of the relevant information, most likely. Any other files with more than one line of logs could also be helpful. Hopefully that will shed some light on this.

Note: I did notice that there is some slight stuttering with AV1 decode at 8K60 on my machine as well during testing my testing of the patch, but of course it's not a regression vs CPU decoding where it will just freeze. It is definitely worth some more investigation at some point, though it'll probably be tricky to nail down the cause. There's also a patch to add a no-copy feature to the WMF implementation which may improve that, but it appears to cause some visual artifacts on my machine.

(In reply to Zaggy1024 from comment #7)

Odd, I can't seem to reproduce this on the beta branch, AV1 decoding works correctly for me on my Intel Xe laptop. Are you able to run some logs for me? If you run Firefox from the command line with args like this:

firefox -MOZ_LOG="MediaSource:5,MediaFormatReader:5,MediaDemuxer:5,PlatformDecoderModule:5,MediaDecoder:5,sync,timestamp" -MOZ_LOG_FILE="%userprofile%\Documents\firefox_log"

you should have some files in your documents folder, firefox_log.child-1.moz_log should be the GPU process, which would hold most of the relevant information, most likely. Any other files with more than one line of logs could also be helpful. Hopefully that will shed some light on this.

Note: I did notice that there is some slight stuttering with AV1 decode at 8K60 on my machine as well during testing my testing of the patch, but of course it's not a regression vs CPU decoding where it will just freeze. It is definitely worth some more investigation at some point, though it'll probably be tricky to nail down the cause. There's also a patch to add a no-copy feature to the WMF implementation which may improve that, but it appears to cause some visual artifacts on my machine.

I'm passing those arguments into the command line, but it seems to generate an empty log file. Am I doing something wrong?

Looks like this is probably a duplicate of Bug 1762125 actually, you have a driver that fits the bill for the bad blocklist entry. Forcing hardware decoding on should be fine for now.

I'm not sure why the logs would fail, it did work on my end. Running Firefox from the command line should reduce the sandboxing so that the processes can write to the log files, but perhaps that wasn't working on your end.

So after reading though the comments on Bug 1762125, it seems that certain Intel drivers are being incorrectly put into a block list? If so, then hopefully a fix should be there soon.

splitting to https://bugzilla.mozilla.org/show_bug.cgi?id=1763680 as requested

I was able to generate the log. The issue I had is outlined here https://bugzilla.mozilla.org/show_bug.cgi?id=1364842#c4. Had to use nightly.

Sorry, wrong thread....

The severity field is not set for this bug.
:jimm, could you have a look please?

For more information, please visit auto_nag documentation.

Forgot to mark this as duplicate after it was confirmed in the other bug, my bad.