All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=827910

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=8a1e2a3537cd0f13f0033fa9ab8fb895c42f4e1541ca6f94c045d1973ef8451e


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

android-one

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/16ab7d1f440000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/16ab7d1f440000

Move all debug print of fetch_benchmark_deps to log print & supress them by default by nednguyen@google.com
https://chromium.googlesource.com/chromium/src/+/8e2a2162848a0791eaf874748e291870dfdf1e17

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

I highly doubt that this bisect is wrong. I just change the how we fetch dependencies in tools/perf/ directory

Let me try running one with a revert of the CL to confirm.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12f1f3a0c40000

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/12f1f3a0c40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14a7e3bf440000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/14a7e3bf440000

The revert didn't reproduce. I'm now going to try re-running the original job. Because of the perf builder migration (issue 792780), the revisions will all be re-built, so we might get a different result.