All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=838455

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=ef1a1a693a162c22e29df4e53e04d802fdc528e644a9896330bbf3bf4224626a


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

android-webview-nexus6
chromium-rel-mac11-pro
chromium-rel-mac12
chromium-rel-win7-gpu-intel

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/15f5c08dc40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1488d553c40000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/1488d553c40000

[ios] make recent tabs disclosure view point up and down by thegreenfrog@chromium.org
https://chromium.googlesource.com/chromium/src/+/7057986ca9e3de962e1059f63ea1942741b1ce3f

[Omnibox] Fetch and draw entity images by dschuyler@chromium.org
https://chromium.googlesource.com/chromium/src/+/c7affc0317e0e19db6af29707cf68c012e17e145

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

hablich, you're listed as the owner of speedometer. Can you help us understand what it measures so we can understand whether and how these pinpointed changes may be related?

The first only affects iOS so it almost certainly isn't involved. The second I'm pretty sure doesn't touch Android and generally doesn't do anything without an experimental flag enabled though it's possible we overlooked something.

But I'm not really clear on whether speedometer measures some aspect of UI performance or startup speed or what.

Speedometer measures JS framework performance and renderer responsiveness. It is very sensitive to V8 performance and Bindings/DOM performance. 

https://browserbench.org/Speedometer/ is the one that is regressing. There is also a revised version.

Seems like it recovered again though. Potentially this was only a bot flake?

Thanks for the explanation, hablich. I don't think either of the culprit CLs would be expected to have an effect on JS or rendered performance.

But out of curiosity, what are you looking at where the graph recovered? Most still look elevated to me at https://chromeperf.appspot.com/group_report?bug_id=838455.

I was looking at the pinpoint result in https://pinpoint-dot-chromeperf.appspot.com/job/1488d553c40000. Seems you are right, the graph has indeed not recovered. 

#8 fyi, that link just gives me an error 500.

hablich@, do you foresee a way that a change in the Omnibox (the address bar UI outside of the HTML content) would cause a performance regression for Speedometer?

Not really, especially as we are running the test in the content shell AFAIK anyway.

It looks like the graphs mentioned in #1 have recovered. 

hablich@ is this still an issue?

Nope, thanks for investigating.