Light up Utf8Utility.*.cs and Utf16Utility.*.cs with Vector512 code paths.

[DeepakRajendrakumaran]

This task is to identify potential opportunities to use Vector512 in these libraries(ASCII/UTF) and add Vector512 paths where possible to further accelerate using SIMD.

@dotnet/avx512-contrib

See below for list of functions which have SIMD paths in them

• GetPointerToFirstInvalidChar(utf16Utility.Validation.cs)
  Note from Ruihan - The existing implementation has Sse2 code path and Vector128 code path, a Sse2.AddSaturate differentiates these 2 code path. We worked internally to upgrade it with Vector256/Avx2, and Vector512/Avx512 code paths, the performance for Vector256 path was fine but the upgrade was blocked by the performance issue we found on Vector512 path due to the execution unit contention between AddSaturate(VPADDUSW) and ExtractMostSignificantBits(VPMOVB2M), which leads to no performance gain compared against Vector256.

• GetPointerToFirstInvalidByte(Utf8Utility.Validation.cs)
  Up for grabs for community to work on

• TranscodeToUtf8(Utf8Utility.Transcoding.cs)
  Up for grabs for community to work on

• AllBytesInVector128AreAscii(Utf8Utility.cs) , Vector128OrdinalIgnoreCaseAscii(Utf8Utility.cs)-> EqualsIgnoreCaseUtf8_Vector128(Ordinal.Utf8.cs) -> EqualsIgnoreCaseUtf8(Ordinal.Utf8.cs)
  Up for grabs for community to work on

[ghost]

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

Issue Details

---

This task is to identify potential opportunities to use Vector512 in these libraries(ASCII/UTF) and add Vector512 paths where possible to further accelerate using SIMD.

@dotnet/avx512-contrib

Author:	DeepakRajendrakumaran
Assignees:	-
Labels:	area-Meta
Milestone:	-

[DeepakRajendrakumaran]

@BruceForstall - Can you please link this to #77034

[ghost]

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

Issue Details

---

This task is to identify potential opportunities to use Vector512 in these libraries(ASCII/UTF) and add Vector512 paths where possible to further accelerate using SIMD.

@dotnet/avx512-contrib

Author:	DeepakRajendrakumaran
Assignees:	-
Labels:	area-Meta, area-System.Text.Encoding, untriaged
Milestone:	-

[BruceForstall]

@BruceForstall - Can you please link this to #77034

Done. Thanks.

[anthonycanino]

@BruceForstall to better track work we did/plan to do, I have made #89280 which covers the upgrades to Ascii.Utility which some of UTF8 encoding uses.

For this issue, I will go through and identify the upgrades that pertain directly to UTF8.Utility and UTF16.Utility, which will be for .NET 9 release.

[Ruihan-Yin]

For GetPointerToFirstInvalidChar in utf16Utility.Validation.cs,

The existing implementation has Sse2 code path and Vector128 code path, a Sse2.AddSaturate differentiates these 2 code path. We worked internally to upgrade it with Vector256/Avx2, and Vector512/Avx512 code paths, the performance for Vector256 path was fine but the upgrade was blocked by the performance issue we found on Vector512 path due to the execution unit contention between AddSaturate(VPADDUSW) and ExtractMostSignificantBits(VPMOVB2M), which leads to no performance gain compared against Vector256.