rc1: Performance regressions in CastingPerf benchmarks

[EgorBo]

Various CastingPerf benchmarks regressed in RC1, e.g.

Namely:

• ObjFooIsObj2
• ObjFooIsObj
• IntObj
• IFooFooIsIFoo
• FooObjIsNull
• FooObjIsFoo2
• FooObjIsFoo
• FooObjCastIfIsa

Commit range: f58a790...01e2521

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.

[EgorBo]

@jakobbotsch can it be #73562? the regression exists only on x64

Another candidates are GC-related changes: #73375 and #67389

[jakobbotsch]

@jakobbotsch can it be #73562? the regression exists only on x64

I don't see how, that change should barely affect x64.

Tagging subscribers to this area: @dotnet/gc
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Various CastingPerf benchmarks regressed in RC1, e.g.

Namely:

• ObjFooIsObj2
• ObjFooIsObj
• IntObj
• IFooFooIsIFoo
• FooObjIsNull
• FooObjIsFoo2
• FooObjIsFoo
• FooObjCastIfIsa

Commit range: f58a790...01e2521

Author:	EgorBo
Assignees:	-
Labels:	tenet-performance, area-GC-coreclr, untriaged
Milestone:	-

[mangod9]

I have verified that setting COMPlus_GCWriteBarrier=3 improves the perf here for wks. Since the precise write barrier has shown better perf for real world scenarios, we will continue to investigate this in 8.