Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[API Proposal]: Arm64: FEAT_F64MM #94025

Open
a74nh opened this issue Oct 26, 2023 · 7 comments
Open

[API Proposal]: Arm64: FEAT_F64MM #94025

a74nh opened this issue Oct 26, 2023 · 7 comments
Labels
api-suggestion Early API idea and discussion, it is NOT ready for implementation arch-arm64 area-System.Runtime.Intrinsics
Milestone
Future

Comments

@a74nh
Copy link
Contributor

a74nh commented Oct 26, 2023 •
edited
Loading 

namespace System.Runtime.Intrinsics.Arm

/// VectorT Summary
public abstract class SveF64mm : AdvSimd /// Feature: FEAT_F64MM
{

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> ConcatenateEvenInt128FromTwoInputs(Vector<T> left, Vector<T> right); // UZP1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> ConcatenateOddInt128FromTwoInputs(Vector<T> left, Vector<T> right); // UZP2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveEvenInt128FromTwoInputs(Vector<T> left, Vector<T> right); // TRN1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<T> left, Vector<T> right); // ZIP2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<T> left, Vector<T> right); // ZIP1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveOddInt128FromTwoInputs(Vector<T> left, Vector<T> right); // TRN2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> LoadVector256AndReplicateToVector(Vector<T> mask, const T *base); // LD1ROH or LD1ROW or LD1ROD or LD1ROB

  public static unsafe Vector<double> MatrixMultiplyAccumulate(Vector<double> op1, Vector<double> op2, Vector<double> op3); // FMMLA // MOVPRFX

  /// total method signatures: 8

}
@ghost
Copy link

ghost commented Oct 26, 2023

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

Issue Details 
namespace System.Runtime.Intrinsics.Arm

/// VectorT Summary
public abstract class Sve : AdvSimd /// Feature: FEAT_F64MM
{
  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> ConcatenateEvenQuadwordsFromTwoInputs(Vector<T> left, Vector<T> right); // UZP1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> ConcatenateOddQuadwordsFromTwoInputs(Vector<T> left, Vector<T> right); // UZP2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveEvenQuadwordsFromTwoInputs(Vector<T> left, Vector<T> right); // TRN1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveOddQuadwordsFromTwoInputs(Vector<T> left, Vector<T> right); // TRN2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveQuadwordsFromHighHalvesOfTwoInputs(Vector<T> left, Vector<T> right); // ZIP2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveQuadwordsFromLowHalvesOfTwoInputs(Vector<T> left, Vector<T> right); // ZIP1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> Load256ReplicateToVector(const T *base); // LD1ROH or LD1ROW or LD1ROD or LD1ROB

  public static unsafe Vector<double> MatrixMultiplyAccumulate(Vector<double> op1, Vector<double> op2, Vector<double> op3);

  /// total method signatures: 8
}
Author: a74nh
Assignees: -
Labels:

area-System.Numerics
Milestone: -

@a74nh
Copy link
Contributor Author

a74nh commented Oct 26, 2023 •
edited
Loading 

/// Full API
public abstract class SveF64mm : AdvSimd /// Feature: FEAT_F64MM
{
    /// ConcatenateEvenInt128FromTwoInputs : Concatenate even quadwords from two inputs

    /// svbfloat16_t svuzp1q[_bf16](svbfloat16_t op1, svbfloat16_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<bfloat16> ConcatenateEvenInt128FromTwoInputs(Vector<bfloat16> left, Vector<bfloat16> right);

    /// svfloat16_t svuzp1q[_f16](svfloat16_t op1, svfloat16_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<half> ConcatenateEvenInt128FromTwoInputs(Vector<half> left, Vector<half> right);

    /// svfloat32_t svuzp1q[_f32](svfloat32_t op1, svfloat32_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<float> ConcatenateEvenInt128FromTwoInputs(Vector<float> left, Vector<float> right);

    /// svfloat64_t svuzp1q[_f64](svfloat64_t op1, svfloat64_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<double> ConcatenateEvenInt128FromTwoInputs(Vector<double> left, Vector<double> right);

    /// svint8_t svuzp1q[_s8](svint8_t op1, svint8_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<sbyte> ConcatenateEvenInt128FromTwoInputs(Vector<sbyte> left, Vector<sbyte> right);

    /// svint16_t svuzp1q[_s16](svint16_t op1, svint16_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<short> ConcatenateEvenInt128FromTwoInputs(Vector<short> left, Vector<short> right);

    /// svint32_t svuzp1q[_s32](svint32_t op1, svint32_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<int> ConcatenateEvenInt128FromTwoInputs(Vector<int> left, Vector<int> right);

    /// svint64_t svuzp1q[_s64](svint64_t op1, svint64_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<long> ConcatenateEvenInt128FromTwoInputs(Vector<long> left, Vector<long> right);

    /// svuint8_t svuzp1q[_u8](svuint8_t op1, svuint8_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<byte> ConcatenateEvenInt128FromTwoInputs(Vector<byte> left, Vector<byte> right);

    /// svuint16_t svuzp1q[_u16](svuint16_t op1, svuint16_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ushort> ConcatenateEvenInt128FromTwoInputs(Vector<ushort> left, Vector<ushort> right);

    /// svuint32_t svuzp1q[_u32](svuint32_t op1, svuint32_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<uint> ConcatenateEvenInt128FromTwoInputs(Vector<uint> left, Vector<uint> right);

    /// svuint64_t svuzp1q[_u64](svuint64_t op1, svuint64_t op2) : "UZP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ulong> ConcatenateEvenInt128FromTwoInputs(Vector<ulong> left, Vector<ulong> right);


    /// ConcatenateOddInt128FromTwoInputs : Concatenate odd quadwords from two inputs

    /// svbfloat16_t svuzp2q[_bf16](svbfloat16_t op1, svbfloat16_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<bfloat16> ConcatenateOddInt128FromTwoInputs(Vector<bfloat16> left, Vector<bfloat16> right);

    /// svfloat16_t svuzp2q[_f16](svfloat16_t op1, svfloat16_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<half> ConcatenateOddInt128FromTwoInputs(Vector<half> left, Vector<half> right);

    /// svfloat32_t svuzp2q[_f32](svfloat32_t op1, svfloat32_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<float> ConcatenateOddInt128FromTwoInputs(Vector<float> left, Vector<float> right);

    /// svfloat64_t svuzp2q[_f64](svfloat64_t op1, svfloat64_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<double> ConcatenateOddInt128FromTwoInputs(Vector<double> left, Vector<double> right);

    /// svint8_t svuzp2q[_s8](svint8_t op1, svint8_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<sbyte> ConcatenateOddInt128FromTwoInputs(Vector<sbyte> left, Vector<sbyte> right);

    /// svint16_t svuzp2q[_s16](svint16_t op1, svint16_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<short> ConcatenateOddInt128FromTwoInputs(Vector<short> left, Vector<short> right);

    /// svint32_t svuzp2q[_s32](svint32_t op1, svint32_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<int> ConcatenateOddInt128FromTwoInputs(Vector<int> left, Vector<int> right);

    /// svint64_t svuzp2q[_s64](svint64_t op1, svint64_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<long> ConcatenateOddInt128FromTwoInputs(Vector<long> left, Vector<long> right);

    /// svuint8_t svuzp2q[_u8](svuint8_t op1, svuint8_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<byte> ConcatenateOddInt128FromTwoInputs(Vector<byte> left, Vector<byte> right);

    /// svuint16_t svuzp2q[_u16](svuint16_t op1, svuint16_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ushort> ConcatenateOddInt128FromTwoInputs(Vector<ushort> left, Vector<ushort> right);

    /// svuint32_t svuzp2q[_u32](svuint32_t op1, svuint32_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<uint> ConcatenateOddInt128FromTwoInputs(Vector<uint> left, Vector<uint> right);

    /// svuint64_t svuzp2q[_u64](svuint64_t op1, svuint64_t op2) : "UZP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ulong> ConcatenateOddInt128FromTwoInputs(Vector<ulong> left, Vector<ulong> right);


    /// InterleaveEvenInt128FromTwoInputs : Interleave even quadwords from two inputs

    /// svbfloat16_t svtrn1q[_bf16](svbfloat16_t op1, svbfloat16_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<bfloat16> InterleaveEvenInt128FromTwoInputs(Vector<bfloat16> left, Vector<bfloat16> right);

    /// svfloat16_t svtrn1q[_f16](svfloat16_t op1, svfloat16_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<half> InterleaveEvenInt128FromTwoInputs(Vector<half> left, Vector<half> right);

    /// svfloat32_t svtrn1q[_f32](svfloat32_t op1, svfloat32_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<float> InterleaveEvenInt128FromTwoInputs(Vector<float> left, Vector<float> right);

    /// svfloat64_t svtrn1q[_f64](svfloat64_t op1, svfloat64_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<double> InterleaveEvenInt128FromTwoInputs(Vector<double> left, Vector<double> right);

    /// svint8_t svtrn1q[_s8](svint8_t op1, svint8_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<sbyte> InterleaveEvenInt128FromTwoInputs(Vector<sbyte> left, Vector<sbyte> right);

    /// svint16_t svtrn1q[_s16](svint16_t op1, svint16_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<short> InterleaveEvenInt128FromTwoInputs(Vector<short> left, Vector<short> right);

    /// svint32_t svtrn1q[_s32](svint32_t op1, svint32_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<int> InterleaveEvenInt128FromTwoInputs(Vector<int> left, Vector<int> right);

    /// svint64_t svtrn1q[_s64](svint64_t op1, svint64_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<long> InterleaveEvenInt128FromTwoInputs(Vector<long> left, Vector<long> right);

    /// svuint8_t svtrn1q[_u8](svuint8_t op1, svuint8_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<byte> InterleaveEvenInt128FromTwoInputs(Vector<byte> left, Vector<byte> right);

    /// svuint16_t svtrn1q[_u16](svuint16_t op1, svuint16_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ushort> InterleaveEvenInt128FromTwoInputs(Vector<ushort> left, Vector<ushort> right);

    /// svuint32_t svtrn1q[_u32](svuint32_t op1, svuint32_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<uint> InterleaveEvenInt128FromTwoInputs(Vector<uint> left, Vector<uint> right);

    /// svuint64_t svtrn1q[_u64](svuint64_t op1, svuint64_t op2) : "TRN1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ulong> InterleaveEvenInt128FromTwoInputs(Vector<ulong> left, Vector<ulong> right);


    /// InterleaveInt128FromHighHalvesOfTwoInputs : Interleave quadwords from high halves of two inputs

    /// svbfloat16_t svzip2q[_bf16](svbfloat16_t op1, svbfloat16_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<bfloat16> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<bfloat16> left, Vector<bfloat16> right);

    /// svfloat16_t svzip2q[_f16](svfloat16_t op1, svfloat16_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<half> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<half> left, Vector<half> right);

    /// svfloat32_t svzip2q[_f32](svfloat32_t op1, svfloat32_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<float> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<float> left, Vector<float> right);

    /// svfloat64_t svzip2q[_f64](svfloat64_t op1, svfloat64_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<double> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<double> left, Vector<double> right);

    /// svint8_t svzip2q[_s8](svint8_t op1, svint8_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<sbyte> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<sbyte> left, Vector<sbyte> right);

    /// svint16_t svzip2q[_s16](svint16_t op1, svint16_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<short> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<short> left, Vector<short> right);

    /// svint32_t svzip2q[_s32](svint32_t op1, svint32_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<int> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<int> left, Vector<int> right);

    /// svint64_t svzip2q[_s64](svint64_t op1, svint64_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<long> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<long> left, Vector<long> right);

    /// svuint8_t svzip2q[_u8](svuint8_t op1, svuint8_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<byte> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<byte> left, Vector<byte> right);

    /// svuint16_t svzip2q[_u16](svuint16_t op1, svuint16_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ushort> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<ushort> left, Vector<ushort> right);

    /// svuint32_t svzip2q[_u32](svuint32_t op1, svuint32_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<uint> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<uint> left, Vector<uint> right);

    /// svuint64_t svzip2q[_u64](svuint64_t op1, svuint64_t op2) : "ZIP2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ulong> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<ulong> left, Vector<ulong> right);


    /// InterleaveInt128FromLowHalvesOfTwoInputs : Interleave quadwords from low halves of two inputs

    /// svbfloat16_t svzip1q[_bf16](svbfloat16_t op1, svbfloat16_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<bfloat16> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<bfloat16> left, Vector<bfloat16> right);

    /// svfloat16_t svzip1q[_f16](svfloat16_t op1, svfloat16_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<half> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<half> left, Vector<half> right);

    /// svfloat32_t svzip1q[_f32](svfloat32_t op1, svfloat32_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<float> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<float> left, Vector<float> right);

    /// svfloat64_t svzip1q[_f64](svfloat64_t op1, svfloat64_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<double> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<double> left, Vector<double> right);

    /// svint8_t svzip1q[_s8](svint8_t op1, svint8_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<sbyte> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<sbyte> left, Vector<sbyte> right);

    /// svint16_t svzip1q[_s16](svint16_t op1, svint16_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<short> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<short> left, Vector<short> right);

    /// svint32_t svzip1q[_s32](svint32_t op1, svint32_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<int> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<int> left, Vector<int> right);

    /// svint64_t svzip1q[_s64](svint64_t op1, svint64_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<long> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<long> left, Vector<long> right);

    /// svuint8_t svzip1q[_u8](svuint8_t op1, svuint8_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<byte> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<byte> left, Vector<byte> right);

    /// svuint16_t svzip1q[_u16](svuint16_t op1, svuint16_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ushort> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<ushort> left, Vector<ushort> right);

    /// svuint32_t svzip1q[_u32](svuint32_t op1, svuint32_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<uint> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<uint> left, Vector<uint> right);

    /// svuint64_t svzip1q[_u64](svuint64_t op1, svuint64_t op2) : "ZIP1 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ulong> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<ulong> left, Vector<ulong> right);


    /// InterleaveOddInt128FromTwoInputs : Interleave odd quadwords from two inputs

    /// svbfloat16_t svtrn2q[_bf16](svbfloat16_t op1, svbfloat16_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<bfloat16> InterleaveOddInt128FromTwoInputs(Vector<bfloat16> left, Vector<bfloat16> right);

    /// svfloat16_t svtrn2q[_f16](svfloat16_t op1, svfloat16_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<half> InterleaveOddInt128FromTwoInputs(Vector<half> left, Vector<half> right);

    /// svfloat32_t svtrn2q[_f32](svfloat32_t op1, svfloat32_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<float> InterleaveOddInt128FromTwoInputs(Vector<float> left, Vector<float> right);

    /// svfloat64_t svtrn2q[_f64](svfloat64_t op1, svfloat64_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<double> InterleaveOddInt128FromTwoInputs(Vector<double> left, Vector<double> right);

    /// svint8_t svtrn2q[_s8](svint8_t op1, svint8_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<sbyte> InterleaveOddInt128FromTwoInputs(Vector<sbyte> left, Vector<sbyte> right);

    /// svint16_t svtrn2q[_s16](svint16_t op1, svint16_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<short> InterleaveOddInt128FromTwoInputs(Vector<short> left, Vector<short> right);

    /// svint32_t svtrn2q[_s32](svint32_t op1, svint32_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<int> InterleaveOddInt128FromTwoInputs(Vector<int> left, Vector<int> right);

    /// svint64_t svtrn2q[_s64](svint64_t op1, svint64_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<long> InterleaveOddInt128FromTwoInputs(Vector<long> left, Vector<long> right);

    /// svuint8_t svtrn2q[_u8](svuint8_t op1, svuint8_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<byte> InterleaveOddInt128FromTwoInputs(Vector<byte> left, Vector<byte> right);

    /// svuint16_t svtrn2q[_u16](svuint16_t op1, svuint16_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ushort> InterleaveOddInt128FromTwoInputs(Vector<ushort> left, Vector<ushort> right);

    /// svuint32_t svtrn2q[_u32](svuint32_t op1, svuint32_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<uint> InterleaveOddInt128FromTwoInputs(Vector<uint> left, Vector<uint> right);

    /// svuint64_t svtrn2q[_u64](svuint64_t op1, svuint64_t op2) : "TRN2 Zresult.Q, Zop1.Q, Zop2.Q"
  public static unsafe Vector<ulong> InterleaveOddInt128FromTwoInputs(Vector<ulong> left, Vector<ulong> right);


    /// LoadVector256AndReplicateToVector : Load and replicate 256 bits of data

    /// svbfloat16_t svld1ro[_bf16](svbool_t pg, const bfloat16_t *base) : "LD1ROH Zresult.H, Pg/Z, [Xarray, Xindex, LSL #1]" or "LD1ROH Zresult.H, Pg/Z, [Xarray, #index * 2]" or "LD1ROH Zresult.H, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<bfloat16> LoadVector256AndReplicateToVector(Vector<bfloat16> mask, const bfloat16 *base);

    /// svfloat16_t svld1ro[_f16](svbool_t pg, const float16_t *base) : "LD1ROH Zresult.H, Pg/Z, [Xarray, Xindex, LSL #1]" or "LD1ROH Zresult.H, Pg/Z, [Xarray, #index * 2]" or "LD1ROH Zresult.H, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<half> LoadVector256AndReplicateToVector(Vector<half> mask, const half *base);

    /// svfloat32_t svld1ro[_f32](svbool_t pg, const float32_t *base) : "LD1ROW Zresult.S, Pg/Z, [Xarray, Xindex, LSL #2]" or "LD1ROW Zresult.S, Pg/Z, [Xarray, #index * 4]" or "LD1ROW Zresult.S, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<float> LoadVector256AndReplicateToVector(Vector<float> mask, const float *base);

    /// svfloat64_t svld1ro[_f64](svbool_t pg, const float64_t *base) : "LD1ROD Zresult.D, Pg/Z, [Xarray, Xindex, LSL #3]" or "LD1ROD Zresult.D, Pg/Z, [Xarray, #index * 8]" or "LD1ROD Zresult.D, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<double> LoadVector256AndReplicateToVector(Vector<double> mask, const double *base);

    /// svint8_t svld1ro[_s8](svbool_t pg, const int8_t *base) : "LD1ROB Zresult.B, Pg/Z, [Xarray, Xindex]" or "LD1ROB Zresult.B, Pg/Z, [Xarray, #index]" or "LD1ROB Zresult.B, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<sbyte> LoadVector256AndReplicateToVector(Vector<sbyte> mask, const sbyte *base);

    /// svint16_t svld1ro[_s16](svbool_t pg, const int16_t *base) : "LD1ROH Zresult.H, Pg/Z, [Xarray, Xindex, LSL #1]" or "LD1ROH Zresult.H, Pg/Z, [Xarray, #index * 2]" or "LD1ROH Zresult.H, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<short> LoadVector256AndReplicateToVector(Vector<short> mask, const short *base);

    /// svint32_t svld1ro[_s32](svbool_t pg, const int32_t *base) : "LD1ROW Zresult.S, Pg/Z, [Xarray, Xindex, LSL #2]" or "LD1ROW Zresult.S, Pg/Z, [Xarray, #index * 4]" or "LD1ROW Zresult.S, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<int> LoadVector256AndReplicateToVector(Vector<int> mask, const int *base);

    /// svint64_t svld1ro[_s64](svbool_t pg, const int64_t *base) : "LD1ROD Zresult.D, Pg/Z, [Xarray, Xindex, LSL #3]" or "LD1ROD Zresult.D, Pg/Z, [Xarray, #index * 8]" or "LD1ROD Zresult.D, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<long> LoadVector256AndReplicateToVector(Vector<long> mask, const long *base);

    /// svuint8_t svld1ro[_u8](svbool_t pg, const uint8_t *base) : "LD1ROB Zresult.B, Pg/Z, [Xarray, Xindex]" or "LD1ROB Zresult.B, Pg/Z, [Xarray, #index]" or "LD1ROB Zresult.B, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<byte> LoadVector256AndReplicateToVector(Vector<byte> mask, const byte *base);

    /// svuint16_t svld1ro[_u16](svbool_t pg, const uint16_t *base) : "LD1ROH Zresult.H, Pg/Z, [Xarray, Xindex, LSL #1]" or "LD1ROH Zresult.H, Pg/Z, [Xarray, #index * 2]" or "LD1ROH Zresult.H, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<ushort> LoadVector256AndReplicateToVector(Vector<ushort> mask, const ushort *base);

    /// svuint32_t svld1ro[_u32](svbool_t pg, const uint32_t *base) : "LD1ROW Zresult.S, Pg/Z, [Xarray, Xindex, LSL #2]" or "LD1ROW Zresult.S, Pg/Z, [Xarray, #index * 4]" or "LD1ROW Zresult.S, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<uint> LoadVector256AndReplicateToVector(Vector<uint> mask, const uint *base);

    /// svuint64_t svld1ro[_u64](svbool_t pg, const uint64_t *base) : "LD1ROD Zresult.D, Pg/Z, [Xarray, Xindex, LSL #3]" or "LD1ROD Zresult.D, Pg/Z, [Xarray, #index * 8]" or "LD1ROD Zresult.D, Pg/Z, [Xbase, #0]"
  public static unsafe Vector<ulong> LoadVector256AndReplicateToVector(Vector<ulong> mask, const ulong *base);


    /// MatrixMultiplyAccumulate : Matrix multiply-accumulate

    /// svfloat64_t svmmla[_f64](svfloat64_t op1, svfloat64_t op2, svfloat64_t op3) : "FMMLA Ztied1.D, Zop2.D, Zop3.D" or "MOVPRFX Zresult, Zop1; FMMLA Zresult.D, Zop2.D, Zop3.D"
  public static unsafe Vector<double> MatrixMultiplyAccumulate(Vector<double> op1, Vector<double> op2, Vector<double> op3);


  /// total method signatures: 85
  /// total method names:      8
}


  /// Total ACLE covered across API:      85

@a74nh
Copy link
Contributor Author

a74nh commented Oct 26, 2023 

/// Rejected:
/// None yet

@a74nh
Copy link
Contributor Author

a74nh commented Oct 26, 2023

This contributes to #93095

It covers all of the instructions in FEAT_F64MM. This an optional 8.2 feature but is not yet available in any hardware.

This list was auto generated from the C ACLE for SVE, and is in three parts:

The methods list reduced down to Vector versions. All possible varaints of T are given above the method.
The complete list of all methods. The corresponding ACLE methods and SVE instructions are given above the method.
All rejected ACLE methods. These are methods we have agreed that do not need including in C#.
Where possible, existing C# naming conventions have been matched.

Many of the C functions include predicate argument(s), of type svbool_t as the first argument. These are missing from the C# method. It is expected that the Jit will create predicates where required, or combine with uses of conditionalSelect(). For more discussion see #88140 comment.

@a74nh a74nh mentioned this issue Oct 26, 2023
Closed
31 tasks
@a74nh
Copy link
Contributor Author

a74nh commented Nov 6, 2023

Updated to reflect review comments from other API proposals.

@a74nh a74nh mentioned this issue Nov 13, 2023
Closed
@kunalspathak kunalspathak added the api-ready-for-review API is ready for review, it is NOT ready for implementation label Feb 7, 2024
@ghost
Copy link

ghost commented Feb 8, 2024

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

Issue Details 
namespace System.Runtime.Intrinsics.Arm

/// VectorT Summary
public abstract class SveF64mm : AdvSimd /// Feature: FEAT_F64MM
{

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> ConcatenateEvenInt128FromTwoInputs(Vector<T> left, Vector<T> right); // UZP1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> ConcatenateOddInt128FromTwoInputs(Vector<T> left, Vector<T> right); // UZP2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveEvenInt128FromTwoInputs(Vector<T> left, Vector<T> right); // TRN1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveInt128FromHighHalvesOfTwoInputs(Vector<T> left, Vector<T> right); // ZIP2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveInt128FromLowHalvesOfTwoInputs(Vector<T> left, Vector<T> right); // ZIP1

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> InterleaveOddInt128FromTwoInputs(Vector<T> left, Vector<T> right); // TRN2

  /// T: bfloat16, half, float, double, sbyte, short, int, long, byte, ushort, uint, ulong
  public static unsafe Vector<T> LoadVector256AndReplicateToVector(Vector<T> mask, const T *base); // LD1ROH or LD1ROW or LD1ROD or LD1ROB

  public static unsafe Vector<double> MatrixMultiplyAccumulate(Vector<double> op1, Vector<double> op2, Vector<double> op3); // FMMLA // MOVPRFX

  /// total method signatures: 8

}
Author: a74nh
Assignees: -
Labels:

area-System.Runtime.Intrinsics, api-ready-for-review
Milestone: -

@kunalspathak kunalspathak mentioned this issue Mar 19, 2024
Closed
@kunalspathak kunalspathak added this to the Future milestone May 1, 2024
@tannergooding tannergooding added api-suggestion Early API idea and discussion, it is NOT ready for implementation arch-arm64 and removed api-ready-for-review API is ready for review, it is NOT ready for implementation labels Aug 20, 2024
@a74nh
Copy link
Contributor Author

a74nh commented Aug 21, 2024

This feature is not yet available on any existing Arm hardware. I don't recommend implementing this for .NET10

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
api-suggestion Early API idea and discussion, it is NOT ready for implementation arch-arm64 area-System.Runtime.Intrinsics
Projects
None yet
Milestone
Future
Development

No branches or pull requests
3 participants
@a74nh @tannergooding @kunalspathak