All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=893820

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=8a8be86a9f6f0ecb20fe91f73df5a86e76a1c48d603da52beae41a222d36a964


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/12f32446e40000

📍 Found significant differences after each of 3 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/12f32446e40000

Center an element before triggering autofill. by uwyiming@google.com
https://chromium.googlesource.com/chromium/src/+/a13016d764f3abe6c6dd2e90cd5cbad3435a8a09
mean_frame_time_renderer_compositor: 32.97 → No values

Add fast path for spreading primitive strings. by dhai@google.com
https://chromium.googlesource.com/v8/v8/+/ef2a19a211acd944a3f19dd8af1df8ee3272feb1
mean_frame_time_renderer_compositor: No values → 32.79

[turbofan] Remove branch_load_poisoning flag. by jarin@chromium.org
https://chromium.googlesource.com/v8/v8/+/b048c16b4f5f1db0642497bc2926ff1cc00cb484
mean_frame_time_renderer_compositor: 32.6 → 40.97 (+8.364)

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

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

This must be some anomaly in the benchmark.

Looking at the graphs, turning on full Spectre mitigations improved performance on these benchmarks(!?). This regression simply returns to the values before I turned on the full mitigations.

FYI, the full mitigations landed with https://chromium.googlesource.com/chromium/src/+/5d9b31e40e833ba98cfce1af3843443ecd04e3d3.

Agree that this behavior is weird. Let's see what happens when a revert patch is tested on a separate Pinpoint job. I'll set that up.

Oh, it looks like the change is in the V8 source tree, not Chromium's. Sorry, I'm not sure how to test a revert change with Pinpoint in this case.

 Issue 893865  has been merged into this issue.

Looks like other benchmarks were affected ( bug 893865 ). Seems worth investigating, if only to conclude that a few benchmarks regress while most others improve.

I see the same pattern in the other benchmarks that regressed on this CL - they all improved on full poisoning mitigation (which will always mean slower JavaScript and more memory consumed by v8). I think the benchmarks must be somehow set up so that slower JavaScript will (perhaps only sometimes) improve score on them. 

 Issue 895267  has been merged into this issue.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/10403e89e40000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/10403e89e40000

Rename IdleSpellCheckCallback to IdleSpellCheckController by xiaochengh@chromium.org
https://chromium.googlesource.com/chromium/src/+/223b5eef45dfe9db1fc3b96f8c6c9cf52925caa0
frame_times: No values → 27.88

Mac OOP-D: Enable OOP-D for Mac Waterfall by fsamuel@chromium.org
https://chromium.googlesource.com/chromium/src/+/00ec720decf72b496b1b5bd5bd16bcc85d04ec03
frame_times: 27.62 → 34.37 (+6.75)

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

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