All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=895292

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=176b667d29f745acafb668b28711d48090f234e37db7b81d50b86fba0383f307


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

mac-10_13_laptop_high_end-perf

system_health.memory_desktop - Benchmark documentation link:
  https://bit.ly/system-health-benchmarks

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11e1520ae40000

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/11e1520ae40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/128cad56e40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/128cad56e40000

Bind canvas color management to GPU compositing by zakerinasab@chromium.org
https://chromium.googlesource.com/chromium/src/+/37aab329373b49a9753c0481e22c192f21861edd
memory:chrome:all_processes:reported_by_chrome:gpu:effective_size: 1.357e+07 → 1.447e+07 (+8.918e+05)

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

Benchmark documentation link:
  https://bit.ly/system-health-benchmarks

This CL only introduces a guard for the new color managed canvas behavior which is still not shipped, and also generally should decrease the memory usage by falling back to legacy behavior. I don't think it can be responsible for this regression.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/168492a2e40000

Re-running bisect. Although it sounds to me like this does affect memory? ("generally decrease the memory usage" - so presumably in some cases increase?).

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/17104882e40000

Not sure how. The thing is that I don't know how this memory usage is measured, but if the layout tests are somehow included, which is the only location we have the new API used, then this guard falls back from half float backed canvas to legacy uint8 backed canvas, so it should be only usage reductions, if any.

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/168492a2e40000

Bind canvas color management to GPU compositing by zakerinasab@chromium.org
https://chromium.googlesource.com/chromium/src/+/37aab329373b49a9753c0481e22c192f21861edd
memory:chrome:all_processes:reported_by_chrome:gpu:effective_size: 1.359e+07 → 1.461e+07 (+1.022e+06)

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

Benchmark documentation link:
  https://bit.ly/system-health-benchmarks

📍 Found significant differences after each of 3 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/17104882e40000

Bind canvas color management to GPU compositing by zakerinasab@chromium.org
https://chromium.googlesource.com/chromium/src/+/37aab329373b49a9753c0481e22c192f21861edd
memory:chrome:all_processes:reported_by_chrome:gpu:effective_size: 1.347e+07 → 1.485e+07 (+1.387e+06)

[libFuzzer] Remove extra spaces in README.md by metzman@chromium.org
https://chromium.googlesource.com/chromium/src/+/83e5ba47c20b796d4da8421e39427fcd0ca669b0
memory:chrome:all_processes:reported_by_chrome:gpu:effective_size: 1.485e+07 → 1.45e+07 (-3.588e+05)

Replace IsAfterStartupTask in RulesetService with PostTaskWithTraits. by ericrobinson@chromium.org
https://chromium.googlesource.com/chromium/src/+/36bf5fece1a539ce254e55a51a0be56718c743d3
memory:chrome:all_processes:reported_by_chrome:gpu:effective_size: 1.429e+07 → 1.482e+07 (+5.303e+05)

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

Benchmark documentation link:
  https://bit.ly/system-health-benchmarks

I'm pretty sure the tool is incorrect that my change caused any significant difference in Chrome since the change only removed two spaces from developer documentation.

My CL changed the delay for the RulesetService indexing/loading from a random amount of time to make it a best-effort task.  I haven't looked deeply into how these tests run, but my CL would potentially influence things at Chrome startup (though shouldn't actually mean much, as the same work is being done, just in a different order).  That being said, it falls outside the revision range listed here: 599217.

I'll look deeper into how my change might interact with these tests, but will also keep an eye to see if someone else resolves this in the meantime.

Available, but no owner or component? Please find a component, as no one will ever find this without one.