These metrics will help us understand the remainder of the pepper usage for video decoding and inform choices around VDA refactoring.
Hardware decoding means VideoDecodeAccelerator is used. Software decoding means VideoDecoderShim is being used.
Since PVDH doesn't get the frame resolution at configuration time, we'll need to record a metric for each frame with a new size and bucket that resolution by HW vs SW. Instead of bucketing a raw size we should use the resolution levels that hbengali@ determined for UKM:
~1080p
~1440p
~2160p or higher
~480p or lower
~720p
We might want a ~360p bucket as well. Initially tagging as M70 if it's not too invasive to merge back so we can get data sooner.
Comment 1 by bugdroid1@chromium.org
, Nov 7