test: play real sizes and formats the GPU refuses in the media player - #325
Merged
Merged
Conversation
The hardware decoding tests used 320x240 media only, which needs no padding and no cropping, and VP9 4:4:4 was the one stream a decoder refuses. New media: 1080p H.264 with B-frames and 1080p VP9 (decoders put 1080 rows in a surface of 1088 or more, and the player has to crop them), 4K H.264, H.264 at 1366x768 (cropped on the right as well as at the bottom), and H.264 High 10 and High 4:4:4 Predictive, which no hardware decoder takes. Each is one second at 15 fps, 2.7 MB together, made with FFmpeg as the test's comment describes. Hardware and software must give the same pictures at the same size, and the two formats must play on in software without an error. Run on an NVIDIA GeForce RTX 3080 on Windows (Direct3D 11): 20 tests pass with hardware required.
devopvoid
force-pushed
the
test/hardware-decoding-sizes
branch
from
October 4, 2026 18:52
9e0f328 to
41a9a8a
Compare
devopvoid
added this pull request to stack #327
October 4, 2026 19:10
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #323: the base is its branch, so the diff is this commit only. Retarget to
mainonce #323 is merged.The hardware decoding tests used 320x240 media only, which needs no padding and no cropping, and VP9 4:4:4 was the one stream a decoder refuses.
New media (2.7 MB together, one second at 15 fps each, made with FFmpeg as the test's comment describes): 1080p H.264 with B-frames, 1080p VP9, 4K H.264, H.264 at 1366x768, and H.264 High 10 and High 4:4:4 Predictive.
Run on an NVIDIA GeForce RTX 3080 on Windows (Direct3D 11): all 20 tests of
HardwareDecodingTestpass withwebrtc.test.hardwareDecodingset.A side finding, from a throwaway benchmark that is not part of this change (steady state, 15 fps, noisy synthetic content, one 8 s run per setting): on this machine the D3D11 path saves little CPU. 1080p H.264 took 7.2 to 8.7 ms of CPU per frame in hardware against 8.7 ms in software; 4K took about 26 ms against 12 to 14 ms. The 7 to 8 times of the Apple M2 does not carry over, so the default should stay software on Windows until the readback and conversion are looked at.