What do we measure for other perf tests? Measuring CPU time will omit any delays caused by inter-thread interactions (since it won't count when threads are blocked), but no idea how big a problem this really is in practise. Also, I'm never sure exactly under which conditions CPU time continues to increment while waiting for IO...

The other perf tests do measure just wall time, and they're quite noisy. Maybe we should try measuring both?

In general, I try to make less metrics if possible. But seems like wall-time do capture some important interactions here, hence measuring both sgtm.

+Alex: can you extend the webview-startup metric to include cpu time?

Agreed with Ned. Both seem important here. If we see a significant regression in wall time but not a significant regression in CPU time, that might be an indication that we have some sort of scheduling regression. If both regress, that means that the problem is more in the critical path of our code.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1684132fc5b8a64a28060573ca98ba6899a13df2

commit 1684132fc5b8a64a28060573ca98ba6899a13df2
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Wed Jun 15 05:12:33 2016

Roll src/third_party/catapult/ 298440299..e319924ad (5 commits).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/298440299790..e319924ad45b

$ git log 298440299..e319924ad --date=short --no-merges --format='%ad %ae %s'

BUG= 619218 , 619068 , 619448 

TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2065873004
Cr-Commit-Position: refs/heads/master@{#399847}

[modify] https://crrev.com/1684132fc5b8a64a28060573ca98ba6899a13df2/DEPS

This is fixed, right? (Please reopen if it's not)