All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=886755

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=b8285fa2e2b46a6a61b99c80463ad89518308527333eb00297f104280765a544


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

Win 7 Nvidia GPU Perf

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

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/170c2484e40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/170c2484e40000

[objects] Change String::length field to uint32_t. by bmeurer@chromium.org
https://chromium.googlesource.com/v8/v8/+/c7a0049e1b70e8f455229c8a470df605ea064f70
5.183e+08 → 5.279e+08 (+9.663e+06)

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

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

The CL in question reduces the size of every String and Symbol by one word (on 64-bit architectures). 32-bit architectures are not affected memory-wise. Otherwise there are no memory-related changes in the CL.

So this seems to be another of these cases where GC pressure goes down and thereby V8 effectively consumes more memory. Not sure if the GC team wants to take a look, but marking as WontFix from my side.