All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=818712

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=c339b4b082eb78b7334fbe262bd994fca400040a05d30e569625cf6771ea203a


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

android-nexus6

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/15818254440000

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/15818254440000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14adbb2a440000

📍 Found significant differences after each of 4 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/14adbb2a440000

[builtins] Refactor the StringConstructor builtin by mathias@chromium.org
https://chromium.googlesource.com/v8/v8/+/15e207b300bf7f7b70df83e98cbaf8c0d99a7485

[runtime] Move validity cell from PrototypeInfo to Map. by ishell@chromium.org
https://chromium.googlesource.com/v8/v8/+/40a3e6dcb9bdb5cb2a7f6e29fa8e4de75e538ef6

[runtime] Always store the name in a function's ScopeInfo by cbruni@chromium.org
https://chromium.googlesource.com/v8/v8/+/01488b9c4f1b4c8e2b66494b24d7e7ff8a826860

[ic] Introduce new IC for storing into array literals. by neis@chromium.org
https://chromium.googlesource.com/v8/v8/+/2e2860f74f9c610d5ffb3ee1778d937055d25dea

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

My CL (biggest regression) is expected to temporarily regress memory (with the CL a function's name is stored both on the SharedFunctionInfo and the ScopeInfo). Future CLs will deduplicate the name field and only store it once.

I think the same counts for ishell@'s CL.

The actual regression on washingtonpost is 1.2%, browse_social just seems to have gotten noiser.

Detaching the JSFunction from the Context made up for this regression by large:

https://chromeperf.appspot.com/report?sid=233f8c36f9dd5a69f8f09496fa1341181f9d972def0f6650e06fb99b7fdbf927&start_rev=554384&end_rev=556392

~ 2.28% improvement overall for the later change
~ 1.7% improvement for the later change  vs the 1% regression introduced by my cl on washingtonpost.