Perf/simd byte equality

[official-markus-maurer]

Summary

• Replaces manual byte-by-byte loops with Span.SequenceEqual(...) for hardware-accelerated comparisons throughout hot paths.
• Adds fast-path ReferenceEquals , consistent null/length guards, and inlining hints to minimize overhead.
• Reduces allocation pressure in scan/hash flows and stream/comparison utilities; preserves existing external behavior.

Key Changes

• Byte array equality (SIMD path):

  • Compress/CompressUtils.cs:84 uses b0.AsSpan().SequenceEqual(b1) .
  • FileScanner/FileScan.cs:534 updates ByteArrCompare to b0.AsSpan().SequenceEqual(b1) .
  • RomVaultCore/Utils/ArrByte.cs:45 updates BCompare ; RomVaultCore/Utils/ArrByte.cs:65 updates ECompare .
  • DATReader/Utils/ArrByte.cs:22 updates bCompare ; DATReader/Utils/ArrByte.cs:41 updates nCompare .
  • Compress/SevenZip/Util.cs:82 updates Compare(this byte[] b1, byte[] b2) .
  • CHDlib/Utils/Util.cs:31 updates ByteArrEquals .

• Imports for MemoryExtensions :

  • Compress/SevenZip/Util.cs:1 , DATReader/Utils/ArrByte.cs:1 , CHDlib/Utils/Util.cs:1 include using System; .

• Package support for netstandard2.0 :

  • Added System.Memory references to Compress/Compress.csproj , RomVaultCore/RomVaultCore.csproj.
  • DATReader/DATReader.csproj , CHDlib/CHDSharpLib.csproj , RVZstdSharp/RVZstdSharp.csproj .

• Misc micro-optimizations (earlier commits visible in history):

  • Throttle progress string updates in hashing loop.
  • Switch buffer management to ArrayPool where applicable.
  • Use string.Equals(..., StringComparison.Ordinal) and inline char normalization to avoid per-call allocations.

Performance Impact

• SequenceEqual over Span leverages vectorized comparisons and reduces bounds-check/branch overhead versus manual loops.
• ReferenceEquals prevents work for identical buffers; early null/length exits minimize iteration.
• Reduced temporary allocations in hashing and string-heavy paths lowers GC pressure on large scans.