All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=907107

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=4d3361267efb46746c0bb47b44f3983528098922d74743a02939823612901a5d


Bot(s) for this bug's original alert(s):

mac-10_12_laptop_low_end-perf
mac-10_13_laptop_high_end-perf

rendering.desktop - Benchmark documentation link:
  https://bit.ly/rendering-benchmarks

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/17a5f34be40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/17a5f34be40000

Fix InProcessCommandBuffer handling of buffer presented. by kylechar@chromium.org
https://chromium.googlesource.com/chromium/src/+/d2b1aadb7232344d8a8fc90569ab3df3e3bc7351
frame_times: 18.36 → 22.93 (+4.566)

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/10172eb7e40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1286bd38140000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/10172eb7e40000

Fix InProcessCommandBuffer handling of buffer presented. by kylechar@chromium.org
https://chromium.googlesource.com/chromium/src/+/d2b1aadb7232344d8a8fc90569ab3df3e3bc7351
frame_times: 23.62 → 20.65 (-2.967)

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Benchmark documentation link:
  https://bit.ly/rendering-benchmarks

 Issue 908577  has been merged into this issue.

There are some improvements and some regressions here but I don't think it's actionable. Before https://crrev.com/c/1337638 we were using the end of the GPU swap as the vsync timebase. After we update the vsync timebase the next begin frame is scheduled based on it. I'd expect it to have some impact on frame_times and any metric computed based on frame timing.

I don't think we want to go back to using an incorrect vsync timebase.