Vectorize Forward/Reflected CRC-32 and Forward CRC-64

[bartonjs]

This provides a vector implementation of CRC-32, and forward-only for CRC-64.

Reflected CRC-64 theoretically requires UInt128 to hold the 65-bit constants. That's not "it can't be done", but it's complicating the algorithm, so it's being pushed to a followup.

Intel's isa-l repository does have a reflected CRC-64 that only uses the xmm registers, and seems to be using only ulong for the constants (e.g. rk20: DQ 0x1b1ab00000000001), so it does seem to be possible, but the tooling isn't coming to that conclusion automatically.

Contributes to #124576.

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-io-hashing, @bartonjs, @vcsjones
See info in area-owners.md if you want to be subscribed.

[Copilot]

Pull request overview

This pull request implements vectorized CRC computation for CRC-32 (both forward and reflected variants) and CRC-64 (forward only). The changes refactor the existing vectorization code from the Crc32 and Crc64 classes into the respective ParameterSet classes, enabling vectorization support for custom CRC polynomials.

Changes:

• Introduces abstract base classes (ForwardCrc32, ReflectedCrc32, ForwardCrc64) that handle vectorization using a partial method pattern
• Adds CrcPolynomialHelper class with a UInt640 struct for computing folding and Barrett constants for arbitrary polynomials
• Migrates vectorization logic from Crc32/Crc64 static classes to parameter set implementations

Reviewed changes

Copilot reviewed 18 out of 18 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
CrcPolynomialHelper.cs	New helper class for computing CRC folding/Barrett constants using 640-bit arithmetic
Crc32ParameterSet.cs	Adds ReflectedCrc32 and ForwardCrc32 abstract base classes with partial method hooks for vectorization
Crc32ParameterSet.Vectorized.cs	Implements vectorized CRC-32 computation for both reflected and forward variants
Crc32ParameterSet.WellKnown.cs	Updates IEEE 802.3 and CRC-32C to use new base classes with ARM intrinsics support
Crc32ParameterSet.Table.cs	Updates table-based implementations to inherit from new base classes
Crc32.cs	Removes partial modifier, delegates to parameter set's Update method
Crc32.Vectorized.cs	Deleted - logic moved to ParameterSet
Crc32.Table.cs	Deleted - table moved to ParameterSet.WellKnown
Crc32.Arm.cs	Deleted - ARM intrinsics moved to ParameterSet.WellKnown
Crc64ParameterSet.cs	Adds ForwardCrc64 abstract base class with vectorization support
Crc64ParameterSet.Vectorized.cs	Implements vectorized forward CRC-64 computation
Crc64ParameterSet.WellKnown.cs	Updates ECMA-182 to use ForwardCrc64 base class
Crc64ParameterSet.Table.cs	Updates table-based forward implementation to inherit from ForwardCrc64
Crc64.cs	Removes partial modifier, delegates to parameter set's Update method
Crc64.Vectorized.cs	Deleted - logic moved to ParameterSet
Crc64.Table.cs	Deleted - table moved to ParameterSet.WellKnown
System.IO.Hashing.csproj	Updates file references to reflect deletions and additions
Crc32Tests_ParameterSet_Custom.cs	Adds test driver for forward CRC-32 with non-standard polynomial

[bartonjs]

Perf-wise, it's basically unchanged from the current vectorized implementation for CRC-32/IEEE and CRC-64/ECMA.

BenchmarkDotNet v0.15.8, Windows 11 (10.0.26200.7840/25H2/2025Update/HudsonValley2)
Intel Core i7-8700 CPU 3.20GHz (Max: 3.19GHz) (Coffee Lake), 1 CPU, 12 logical and 6 physical cores
.NET SDK 11.0.100-preview.1.26104.118
  [Host]     : .NET 10.0.3 (10.0.3, 10.0.326.7603), X64 RyuJIT x86-64-v3
  Job-QMWUGV : .NET 11.0.0 (11.0.0-dev, 42.42.42.42424), X64 RyuJIT x86-64-v3
  .NET 10.0  : .NET 10.0.3 (10.0.3, 10.0.326.7603), X64 RyuJIT x86-64-v3

Runtime=.NET 10.0  

Method	Job	Toolchain	size	Mean	Error	StdDev	Allocated
Crc32_Prebuilt	Job-QMWUGV	CoreRun	0	2.5973 ns	0.0234 ns	0.0218 ns	-
Crc32_Prebuilt	.NET 10.0	Default	0	2.3792 ns	0.0178 ns	0.0139 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	0	0.7268 ns	0.0303 ns	0.0283 ns	-
Crc64_Prebuilt	.NET 10.0	Default	0	0.7403 ns	0.0070 ns	0.0062 ns	-
Crc32_Prebuilt	Job-QMWUGV	CoreRun	10	9.6688 ns	0.0179 ns	0.0149 ns	-
Crc32_Prebuilt	.NET 10.0	Default	10	8.8438 ns	0.0230 ns	0.0204 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	10	8.3621 ns	0.0231 ns	0.0205 ns	-
Crc64_Prebuilt	.NET 10.0	Default	10	8.3531 ns	0.0239 ns	0.0223 ns	-
Crc32_Prebuilt	Job-QMWUGV	CoreRun	100	13.3525 ns	0.0356 ns	0.0333 ns	-
Crc32_Prebuilt	.NET 10.0	Default	100	14.3781 ns	0.0298 ns	0.0279 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	100	14.8436 ns	0.0454 ns	0.0403 ns	-
Crc64_Prebuilt	.NET 10.0	Default	100	14.7429 ns	0.0191 ns	0.0149 ns	-
Crc32_Prebuilt	Job-QMWUGV	CoreRun	1000	56.0132 ns	0.0887 ns	0.0740 ns	-
Crc32_Prebuilt	.NET 10.0	Default	1000	55.7654 ns	0.0795 ns	0.0744 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	1000	64.1417 ns	0.0877 ns	0.0778 ns	-
Crc64_Prebuilt	.NET 10.0	Default	1000	63.6507 ns	0.1179 ns	0.1045 ns	-
Crc32_Prebuilt	Job-QMWUGV	CoreRun	10000	347.9694 ns	0.5755 ns	0.5383 ns	-
Crc32_Prebuilt	.NET 10.0	Default	10000	348.6138 ns	0.6414 ns	0.5999 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	10000	444.4825 ns	0.4317 ns	0.4038 ns	-
Crc64_Prebuilt	.NET 10.0	Default	10000	443.5430 ns	0.6813 ns	0.6373 ns	-
Crc32_Prebuilt	Job-QMWUGV	CoreRun	100000	3,444.3479 ns	5.1116 ns	4.2684 ns	-
Crc32_Prebuilt	.NET 10.0	Default	100000	3,444.2575 ns	3.7278 ns	3.1128 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	100000	4,402.8019 ns	8.5030 ns	7.9537 ns	-
Crc64_Prebuilt	.NET 10.0	Default	100000	4,401.7223 ns	7.0760 ns	6.2727 ns	-
Crc32_Prebuilt	Job-QMWUGV	CoreRun	1000000	36,523.5935 ns	296.2984 ns	277.1577 ns	-
Crc32_Prebuilt	.NET 10.0	Default	1000000	36,608.2642 ns	236.4533 ns	221.1786 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	1000000	44,802.2878 ns	72.3593 ns	67.6849 ns	-
Crc64_Prebuilt	.NET 10.0	Default	1000000	44,824.7990 ns	63.7654 ns	56.5264 ns	-
Crc32_Prebuilt	Job-QMWUGV	CoreRun	1048576	38,287.0065 ns	210.2360 ns	186.3687 ns	-
Crc32_Prebuilt	.NET 10.0	Default	1048576	38,268.0418 ns	279.9785 ns	261.8921 ns	-
Crc64_Prebuilt	Job-QMWUGV	CoreRun	1048576	47,016.1375 ns	106.4478 ns	99.5714 ns	-
Crc64_Prebuilt	.NET 10.0	Default	1048576	47,013.3135 ns	74.0468 ns	65.6406 ns	-