Consume new PGO / IBC data from optimization repo

[AndyAyersMS]

We saw a number of recent perf regressions on windows that were traced back to stale PGO data for the runtime.

This seems to have happened as a consequence of #41897, the change to enable builds via ninja; perhaps the reworking of the PCH or similar is the culprit.

You can see from release build logs that very few methods now get optimized for speed; eg:

;; before
2324 of 33506 (  6.94%) profiled functions will be compiled for speed
;; after
 194 of 33461 (  0.58%) profiled functions will be compiled for speed

Not clear if non-Windows builds are as impacted.

I wanted to raise this because (a) recent local builds might see unexpected perf regressions, particularly comparing vs 5.0 RTM; and (b) it is not clear when we might see new PGO data.

[Dotnet-GitSync-Bot]

I couldn't figure out the best area label to add to this issue. If you have write-permissions please help me learn by adding exactly one area label.

[AndyAyersMS]

Not sure of area label either, marking as infrastructure for now.

[jkoritzinsky]

We haven’t ingested new PGO data since early August so we’re running quite a bit behind (in comparison to when we would regularly ingest new data).

[AndyAyersMS]

@brianrob you mentioned offline there were some discussions on who was going to own collection of native PGO data. Who should I nag about resolving this?

[AndyAyersMS]

@jeffschwMSFT maybe you can update this with status on resolving native PGO updates?

[jeffschwMSFT]

I have asked @agocke to fold the upcoming work as part of our infrastructure initiative. One of our first commitments is to produce an instrumented build that can then be profiled in our new system (which is tbd).

[agocke]

Still getting up to speed on how this works, but I hope to have a better idea by the end of the week

[agocke]

This should be fixed now. @DrewScoggins is there any way to look at some benchmarks and verify the results?

[DrewScoggins]

This should give some results on dotnet build, which is the main scenarios we run that should be effected. https://msit.powerbi.com/groups/9ae0c009-7b27-44c9-8360-4452ec48cef3/reports/80d74b6f-53f1-4b44-b355-2913714ee37d/ReportSectionb89e16b5be17519f1bb8