media/gpu/vaapi: batch submitting VABufferIDs for VP8 decoding
This CL follows crrev.com/c/2393629 by batching up SubmitBuffer calls in the VP8 decode accelerator. This is verified via chrome:tracing and codepen.io/full/qzYXba that plays 4 1280x572 VP8 videos at the same time. Tracing is captured for a few seconds, basically the SubmitBuffers() call takes ~0.312ms/call versus the individual SubmitBuffer() calls ~0.072ms -- which gives an aggregate of 0.360ms, or savings of about ~20%. Since decoding itself takes about the same as SubmitBuffers on my BSW device (reks), batching gives ~10% reduction in decode time. Bug: b/166646505 Change-Id: I11c5570bea1c049fe0e9af75239bd60ac906a843 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2401512 Commit-Queue: Miguel Casas <mcasas@chromium.org> Reviewed-by:J Kardatzke <jkardatzke@chromium.org> Cr-Commit-Position: refs/heads/master@{#805556}
Showing
Please register or sign in to comment