All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=874898

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=493507a9c1abb854a55fe241ae33396460d417c6b0ff48c024c1e94290bcb251


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

Android Nexus5 Perf

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14b3e0fa640000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/14b3e0fa640000

Introduced blink::mojom::WebUsbService with restricted methods. by donna.wu@intel.com
https://chromium.googlesource.com/chromium/src/+/838ac3604d917b78db1492344ead40d36cbd2b12
24.99 → 18.55 (-6.438)

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

Reilly@, I looked at the implementation of createImageBitmap() according to the reported test case. But it seems not be related with USB. I supposed that the CL will only affect WebUSB in the renderer. Any hints?

It looks like this benchmark is oscillating between two results on a regular basis (see the graphs at the link above). It is unlikely that Donna's change is related. Reassigning to tdresser@ to look at the overall behavior of this benchmark.

Over to fserb@, who is an owner of this benchmark.

This was just a flakiness on the test.