diff options
| author | hendrikw <hendrikw@chromium.org> | 2015-05-06 13:11:21 -0700 |
|---|---|---|
| committer | Commit bot <commit-bot@chromium.org> | 2015-05-06 20:12:03 +0000 |
| commit | 647a15fe0d892ee01f688df1e90b3c3d12bc7ace (patch) | |
| tree | c944ddde8b348395ec8529f676f8c3feb2ba68c4 /cc/debug/frame_rate_counter.cc | |
| parent | c95d6ff011264d5d6b6524d93adfdea178abf6a0 (diff) | |
| download | chromium_src-647a15fe0d892ee01f688df1e90b3c3d12bc7ace.zip chromium_src-647a15fe0d892ee01f688df1e90b3c3d12bc7ace.tar.gz chromium_src-647a15fe0d892ee01f688df1e90b3c3d12bc7ace.tar.bz2 | |
cc: Make the fps meter lie less
The fps meter measures the delta between frames, so, for example, you may
perform two scrolls, with 1/10 of a second between, and it will report 10fps.
If, you however, took longer than a 1/4 second, that would be filtered out.
As a result, really slow pages would report the last good frame. I'm
increasing the filter time to 1.5s, which kindof sucks.
We could try to measure the actual time frame time, but then the fps meter
would not report correctly in other cases (e.g. an animated gif at 5fps
should report 5fps)
BUG=446589
Review URL: https://codereview.chromium.org/1126303003
Cr-Commit-Position: refs/heads/master@{#328594}
Diffstat (limited to 'cc/debug/frame_rate_counter.cc')
| -rw-r--r-- | cc/debug/frame_rate_counter.cc | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/cc/debug/frame_rate_counter.cc b/cc/debug/frame_rate_counter.cc index 2f3b291..c43270d 100644 --- a/cc/debug/frame_rate_counter.cc +++ b/cc/debug/frame_rate_counter.cc @@ -21,7 +21,7 @@ namespace cc { // - if the frame is too slow, then there is probably not animating content, so // we should not pollute the average. static const double kFrameTooFast = 1.0 / 70.0; -static const double kFrameTooSlow = 1.0 / 4.0; +static const double kFrameTooSlow = 1.5; // If a frame takes longer than this threshold (measured in seconds) then we // (naively) assume that it missed a screen refresh; that is, we dropped a |