Regression in frame build times on iOS

[zanderso]

https://flutter-flutter-perf.skia.org/e/?queries=sub_result%3D90th_percentile_frame_build_time_millis%26sub_result%3D99th_percentile_frame_build_time_millis%26sub_result%3Daverage_frame_build_time_millis%26sub_result%3Dworst_frame_build_time_millis%26test%3D~Mac_ios

In the Engine roll here #92313

That contained only a Dart roll: https://dart.googlesource.com/sdk.git/+log/84c74548e30a..64f7ed6aa45a

Since this is iOS specific, my guess would be that this is the re-land of the CPU profiling change from @bkonyi: https://dart.googlesource.com/sdk.git/+/db45ae178a71c4df41ceec3c28c6d562dedc132f.

The effect is a lot bigger on ios32. However, there also seems to be an impact on worst case frame build times on the 64-bit devices.

[zanderso]

There is also an iOS-specific regression in memory footprint, which I'm guessing is due to the same cause: https://flutter-flutter-perf.skia.org/e/?begin=1634854082&end=1635154528&keys=X475e3980825cc4b5831b0e57c9feeacd&num_commits=50&request_type=1&xbaroffset=26313. @bkonyi Is the size of the regression here in-line with what is expected for the CPU profile sampling buffer?

[bkonyi]

There is also an iOS-specific regression in memory footprint, which I'm guessing is due to the same cause: https://flutter-flutter-perf.skia.org/e/?begin=1634854082&end=1635154528&keys=X475e3980825cc4b5831b0e57c9feeacd&num_commits=50&request_type=1&xbaroffset=26313. @bkonyi Is the size of the regression here in-line with what is expected for the CPU profile sampling buffer?

Yeah, ~5MB is probably in the ballpark for the sample buffer.

[chinmaygarde]

From Triage: Some regression in just profile mode is acceptable (release mode is unaffected) since this enables profiling features that were previously disabled. Since this disproportionately affect 32 bit iOS devices, we are going to reduce the profiling frequency on that target.

On Android, profiling in on by default. So we should strive to match behavior on iOS. Profiling can be toggled (by DevTools) so benchmark author should be able to keep previous results consistent.

[github-actions]

This thread has been automatically locked since there has not been any recent activity after it was closed. If you are still experiencing a similar issue, please open a new bug, including the output of flutter doctor -v and a minimal reproduction of the issue.