All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=822722

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=c5e9cca10807ca52ad1b3fc31fd529b92c56043a8612cea65dc7aa3bb1640b22


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

chromium-rel-mac11-air
chromium-rel-mac11-pro
chromium-rel-mac12

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11988b21440000

Very interesting. This is a combination of an underflow bug, recently fixed: https://chromium-review.googlesource.com/c/chromium/src/+/955908

And what is likely to be a misuse of shared memory on macOS. I'll investigate further.

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/11988b21440000

Revert "[test_env.py] Reland: Warm up vpython virtualenv cache on swarming task shards" by charliea@chromium.org
https://chromium.googlesource.com/chromium/src/+/5908c5d2269594aff9c2454c4732360a311c27ca

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

Explaining the pinpoint result: the X's mean the test was broken when run, and it started passing at the revert at r538843. Looks like the bisect can't find the culprit until https://github.com/catapult-project/catapult/issues/4335 is fixed.

The underflow is fixed here: https://bugs.chromium.org/p/chromium/issues/detail?id=812346#c4

But the underlying issue is that there is a giant discrepancy between resident shared memory and faulted shared memory in the browser process. More info will be posted in  Issue 812346 .