Enable AVX-512 for block unrollings (both copying and zeroing)

[EgorBo]

Closes#83798

This PR enables AVX-512 for various unrollings using GT_BLK - it can be stackalloc zeroing, struct copy/initialization, Unsafe.InitBloc/Unsafe.BlockCopy calls, etc.

Examples:

structMyStruct{longa,b,c,d,e,f,g,h;}// CopyingMyStructStructCopy(MyStructs){returns;}// ZeroingvoidStackallocZeroing(){byte*ptr=stackallocbyte[300];Consume(ptr);}

Codegen diff: https://www.diffchecker.com/cxc6UYLf/ (this PR is on the right)

As the result, it increases ranges where we previously used to fallback to memcpy/memset calls.

Benchmark:

[Benchmark]publicvoidTest(){varptr=stackalloclong[42];Consume(ptr);}[MethodImpl(MethodImplOptions.NoInlining)]staticvoidConsume(void*ptr){}

Method	Job	Toolchain	Mean
Test	Job-FEQJHS	\Core_Root\corerun.exe	4.229 ns
Test	Job-CMMDYR	\Core_Root_PR\corerun.exe	2.262 ns

Ryzen 7950x, avx512, win-x64

[ghost]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Closes #83798

This PR enables AVX-512 for various unrollings using GT_BLK - it can be stackalloc zeroing, struct copy/initialization, Unsafe.InitBloc/Unsafe.BlockCopy calls, etc.

Examples:

structMyStruct{longa,b,c,d,e,f,g,h;}// CopyingMyStructStructCopy(MyStructs){returns;}// ZeroingvoidStackallocZeroing(){byte*ptr=stackallocbyte[300];Consume(ptr);}

old codegen:

; Method Tests:StructCopy(Tests+MyStruct):Tests+MyStruct:thisvzeroupper vmovdqu ymm0, ymmword ptr [r8] vmovdqu ymmword ptr [rdx],ymm0 vmovdqu ymm0, ymmword ptr [r8+20H] vmovdqu ymmword ptr [rdx+20H],ymm0movrax,rdxvzeroupperret; Total bytes of code: 30; Method Tests:StackallocZeroing():thispushrbpsubrsp,48learbp,[rsp+20H]movrax,0xD1FFAB1Emov qword ptr [rbp],raxtest dword ptr [rsp],espsubrsp,304learcx,[rsp+20H]mov qword ptr [rbp+08H],rcxxoredx,edxmovr8d,304call CORINFO_HELP_MEMSETmovrcx, qword ptr [rbp+08H]call[Tests:Consume(ulong)]movrcx,0xD1FFAB1Ecmp qword ptr [rbp],rcxje SHORT G_M16409_IG03call CORINFO_HELP_FAIL_FASTnoplearsp,[rbp+10H]poprbpret; Total bytes of code: 94

new codegen:

; Method Tests:StructCopy(Tests+MyStruct):Tests+MyStruct:thisvzerouppervmovdqu32zmm0, zmmword ptr [r8]vmovdqu32 zmmword ptr [rdx],zmm0movrax,rdxvzeroupperret; Total bytes of code: 22; Method Tests:StackallocZeroing():thispushrbpsubrsp,48vzeroupperlearbp,[rsp+20H]movrax,0xD1FFAB1Emov qword ptr [rbp+08H],raxtest dword ptr [rsp],espsubrsp,304learcx,[rsp+20H] vxorps zmm0,zmm0vmovdqu32 zmmword ptr [rcx],zmm0vmovdqu32 zmmword ptr [rcx+40H],zmm0vmovdqu32 zmmword ptr [rcx+80H],zmm0vmovdqu32 zmmword ptr [rcx+C0H],zmm0vmovdqu32 zmmword ptr [rcx+F0H],zmm0call[Tests:Consume(ulong)]movrcx,0xD1FFAB1Ecmp qword ptr [rbp+08H],rcxje SHORT G_M16409_IG03call CORINFO_HELP_FAIL_FASTnoplearsp,[rbp+10H]poprbpret; Total bytes of code: 119

Author:	EgorBo
Assignees:	-
Labels:	area-CodeGen-coreclr
Milestone:	-

[EgorBo]

/azp list

[EgorBo]

/azp run runtime-coreclr outerloop, runtime-coreclr jitstress-isas-x86

[azure-pipelines]

Azure Pipelines successfully started running 2 pipeline(s).

[EgorBo]

Diffs
Size regressions are expected (call memset/memcpy is smaller)

[EgorBo]

@tannergooding@BruceForstall @dotnet/avx512-contrib PTAL, I didn't enable it for non-zeroing init (e.g. Unsafe.InitBlockUnaligned(ref a, value: 42, count: 32)) because I plan to work on that separately, even for XMM/YMM it might be improved

[EgorBo]

Failures are mostly #85403

[EgorBo]

benchmark:

[Benchmark]publicvoidTest(){varptr=stackalloclong[42];Consume(ptr);}[MethodImpl(MethodImplOptions.NoInlining)]staticvoidConsume(void*ptr){}}

Method	Job	Toolchain	Mean
Test	Job-FEQJHS	\Core_Root\corerun.exe	4.229 ns
Test	Job-CMMDYR	\Core_Root_PR\corerun.exe	2.262 ns