Regression to cold start (main = 50ms, view = 20ms; around 5/10/22)
Our nightly results show a regression in cold main start up on 5/10 of about 50ms:

And a smaller, gradual increase in cold start view around the same time – somewhere between 5/10 and 5/15:

The period before the large jump averages out to 1.92s (eyeballed) while we see 1.944s on 5/14: about a 20ms increase.
There is a temporary spike that starts earlier, on 5/2, that may also be related to the root cause of this regression.
I talked with the team to transfer me a device with a lower android version in order to catch the differences easily, so I will postpone this ticket until then. If someone else can help feel free to take the ticket
@mcomella I was that this regression had a spike only between 05/01 and 05/16 after that it came back to normal . Do you think it is worth investigating? Thank you !
Triage: Close since it does seems inactionable with the noise. The time was ~600-605ms before the regression on first frame and afterwards was ~630-650ms. It would be very difficult to find on a profile.