Edit

Share via

YMM embedded rounding removed from AVX10.2

Support for YMM embedded rounding has been removed from the Avx10v2 type.

Previous behavior

In previous .NET 10 Preview versions, the affected APIs in Avx10v2 were available.

New behavior

Starting in Preview 5, the affected APIs in Avx10v2 are removed and no longer available.

Version introduced

.NET 10 Preview 5

Type of breaking change

This change can affect binary compatibility and source compatibility.

Reason for change

Intel pivoted direction and now requires that AVX10.2 also implement AVX512. Since ZMM embedded rounding is always available, the YMM embedded rounding feature isn't necessary.

Because the hardware isn't yet available, no users should be affected by this change.

Affected APIs

  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToSByteWithSaturationAndZeroExtendToInt32
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToByteWithSaturationAndZeroExtendToInt32
  • System.Runtime.Intrinsics.X86.Avx10v2.Add
  • System.Runtime.Intrinsics.X86.Avx10v2.Divide
  • System.Runtime.Intrinsics.X86.Avx10v2.Multiply
  • System.Runtime.Intrinsics.X86.Avx10v2.Scale
  • System.Runtime.Intrinsics.X86.Avx10v2.Sqrt
  • System.Runtime.Intrinsics.X86.Avx10v2.Subtract
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector128Int32
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector128Single
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector128UInt32
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector256Double
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector256Int32
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector256Int64
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector256Single
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector256UInt32
  • System.Runtime.Intrinsics.X86.Avx10v2.ConvertToVector256UInt64