Arm64/SVE: Implement ConvertToInt32 and ConvertToUInt32 for float (#1…

…03098) * Added ConverToInt32 and ConvertToUInt32 for float inputs. * Added flags to handle only low predicate registers. * Fix whitespace * Remove special codegen flag * Added new test template for operations with different return types. * Add new test template.
dotnet · Jun 12, 2024 · ced2117 · ced2117
1 parent 9a8c69a
commit ced2117
Show file tree

Hide file tree

Showing 7 changed files with 474 additions and 40 deletions.
diff --git a/src/coreclr/jit/hwintrinsiclistarm64sve.h b/src/coreclr/jit/hwintrinsiclistarm64sve.h
@@ -32,6 +32,8 @@ HARDWARE_INTRINSIC(Sve,           Compute16BitAddresses,
 HARDWARE_INTRINSIC(Sve,           Compute32BitAddresses,                                            -1,      2,      true,  {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_sve_adr,        INS_invalid,        INS_sve_adr,        INS_invalid,        INS_invalid},     HW_Category_SIMD,                  HW_Flag_Scalable|HW_Flag_SpecialCodeGen)
 HARDWARE_INTRINSIC(Sve,           Compute64BitAddresses,                                            -1,      2,      true,  {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_sve_adr,        INS_invalid,        INS_sve_adr,        INS_invalid,        INS_invalid},     HW_Category_SIMD,                  HW_Flag_Scalable|HW_Flag_SpecialCodeGen)
 HARDWARE_INTRINSIC(Sve,           ConditionalSelect,                                                -1,      3,      true,  {INS_sve_sel,        INS_sve_sel,        INS_sve_sel,        INS_sve_sel,        INS_sve_sel,        INS_sve_sel,        INS_sve_sel,        INS_sve_sel,        INS_sve_sel,        INS_sve_sel},     HW_Category_SIMD,                  HW_Flag_Scalable|HW_Flag_ExplicitMaskedOperation|HW_Flag_SupportsContainment)
+HARDWARE_INTRINSIC(Sve,           ConvertToInt32,                                                   -1,      -1,     false, {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_sve_fcvtzs,     INS_invalid},     HW_Category_SIMD,                  HW_Flag_Scalable|HW_Flag_BaseTypeFromFirstArg|HW_Flag_EmbeddedMaskedOperation|HW_Flag_LowMaskedOperation)
+HARDWARE_INTRINSIC(Sve,           ConvertToUInt32,                                                  -1,      -1,     false, {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_sve_fcvtzu,     INS_invalid},     HW_Category_SIMD,                  HW_Flag_Scalable|HW_Flag_BaseTypeFromFirstArg|HW_Flag_EmbeddedMaskedOperation|HW_Flag_LowMaskedOperation)
 HARDWARE_INTRINSIC(Sve,           Count16BitElements,                                                0,      1,      false, {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_sve_cnth,       INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Scalar,                HW_Flag_Scalable|HW_Flag_HasEnumOperand|HW_Flag_SpecialCodeGen|HW_Flag_NoFloatingPointUsed)
 HARDWARE_INTRINSIC(Sve,           Count32BitElements,                                                0,      1,      false, {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_sve_cntw,       INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Scalar,                HW_Flag_Scalable|HW_Flag_HasEnumOperand|HW_Flag_SpecialCodeGen|HW_Flag_NoFloatingPointUsed)
 HARDWARE_INTRINSIC(Sve,           Count64BitElements,                                                0,      1,      false, {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_sve_cntd,       INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Scalar,                HW_Flag_Scalable|HW_Flag_HasEnumOperand|HW_Flag_SpecialCodeGen|HW_Flag_NoFloatingPointUsed)

diff --git a/...ries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Arm/Sve.PlatformNotSupported.cs b/...ries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Arm/Sve.PlatformNotSupported.cs
@@ -848,6 +848,30 @@ public new abstract class Arm64 : AdvSimd.Arm64
         public static unsafe Vector<double> ConditionalSelect(Vector<double> mask, Vector<double> left, Vector<double> right) { throw new PlatformNotSupportedException(); }
 
 
+        ///  ConvertToInt32 : Floating-point convert
+
+        /// <summary>
+        /// svint32_t svcvt_s32[_f32]_m(svint32_t inactive, svbool_t pg, svfloat32_t op)
+        ///   FCVTZS Ztied.S, Pg/M, Zop.S
+        /// svint32_t svcvt_s32[_f32]_x(svbool_t pg, svfloat32_t op)
+        ///   FCVTZS Ztied.S, Pg/M, Ztied.S
+        /// svint32_t svcvt_s32[_f32]_z(svbool_t pg, svfloat32_t op)
+        /// </summary>
+        public static unsafe Vector<int> ConvertToInt32(Vector<float> value) { throw new PlatformNotSupportedException(); }
+
+
+        ///  ConvertToUInt32 : Floating-point convert
+
+        /// <summary>
+        /// svuint32_t svcvt_u32[_f32]_m(svuint32_t inactive, svbool_t pg, svfloat32_t op)
+        ///   FCVTZU Ztied.S, Pg/M, Zop.S
+        /// svuint32_t svcvt_u32[_f32]_x(svbool_t pg, svfloat32_t op)
+        ///   FCVTZU Ztied.S, Pg/M, Ztied.S
+        /// svuint32_t svcvt_u32[_f32]_z(svbool_t pg, svfloat32_t op)
+        /// </summary>
+        public static unsafe Vector<uint> ConvertToUInt32(Vector<float> value) { throw new PlatformNotSupportedException(); }
+
+
         ///  Count16BitElements : Count the number of 16-bit elements in a vector
 
         /// <summary>

diff --git a/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Arm/Sve.cs b/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Arm/Sve.cs
@@ -905,6 +905,30 @@ public new abstract class Arm64 : AdvSimd.Arm64
         public static unsafe Vector<double> ConditionalSelect(Vector<double> mask, Vector<double> left, Vector<double> right) => ConditionalSelect(mask, left, right);
 
 
+        ///  ConvertToInt32 : Floating-point convert
+
+        /// <summary>
+        /// svint32_t svcvt_s32[_f32]_m(svint32_t inactive, svbool_t pg, svfloat32_t op)
+        ///   FCVTZS Ztied.S, Pg/M, Zop.S
+        /// svint32_t svcvt_s32[_f32]_x(svbool_t pg, svfloat32_t op)
+        ///   FCVTZS Ztied.S, Pg/M, Ztied.S
+        /// svint32_t svcvt_s32[_f32]_z(svbool_t pg, svfloat32_t op)
+        /// </summary>
+        public static unsafe Vector<int> ConvertToInt32(Vector<float> value) => ConvertToInt32(value);
+
+
+        ///  ConvertToUInt32 : Floating-point convert
+
+        /// <summary>
+        /// svuint32_t svcvt_u32[_f32]_m(svuint32_t inactive, svbool_t pg, svfloat32_t op)
+        ///   FCVTZU Ztied.S, Pg/M, Zop.S
+        /// svuint32_t svcvt_u32[_f32]_x(svbool_t pg, svfloat32_t op)
+        ///   FCVTZU Ztied.S, Pg/M, Ztied.S
+        /// svuint32_t svcvt_u32[_f32]_z(svbool_t pg, svfloat32_t op)
+        /// </summary>
+        public static unsafe Vector<uint> ConvertToUInt32(Vector<float> value) => ConvertToUInt32(value);
+
+
         ///  Count16BitElements : Count the number of 16-bit elements in a vector
 
         /// <summary>

diff --git a/src/libraries/System.Runtime.Intrinsics/ref/System.Runtime.Intrinsics.cs b/src/libraries/System.Runtime.Intrinsics/ref/System.Runtime.Intrinsics.cs
@@ -4307,11 +4307,14 @@ public new abstract partial class Arm64 : System.Runtime.Intrinsics.Arm.AdvSimd.
         public static System.Numerics.Vector<float> ConditionalSelect(System.Numerics.Vector<float> mask, System.Numerics.Vector<float> left, System.Numerics.Vector<float> right) { throw null; }
         public static System.Numerics.Vector<double> ConditionalSelect(System.Numerics.Vector<double> mask, System.Numerics.Vector<double> left, System.Numerics.Vector<double> right) { throw null; }
 
+        public static System.Numerics.Vector<int> ConvertToInt32(System.Numerics.Vector<float> value) { throw null; }
+        public static System.Numerics.Vector<uint> ConvertToUInt32(System.Numerics.Vector<float> value) { throw null; }
+
         public static ulong Count16BitElements([ConstantExpected] SveMaskPattern pattern = SveMaskPattern.All) { throw null; }
         public static ulong Count32BitElements([ConstantExpected] SveMaskPattern pattern = SveMaskPattern.All) { throw null; }
         public static ulong Count64BitElements([ConstantExpected] SveMaskPattern pattern = SveMaskPattern.All) { throw null; }
         public static ulong Count8BitElements([ConstantExpected] SveMaskPattern pattern = SveMaskPattern.All) { throw null; }
-        
+
         public static System.Numerics.Vector<byte> CreateFalseMaskByte() { throw null; }
         public static System.Numerics.Vector<double> CreateFalseMaskDouble() { throw null; }
         public static System.Numerics.Vector<short> CreateFalseMaskInt16() { throw null; }