Skip to content
This repository has been archived by the owner on Jan 23, 2023. It is now read-only.

Improve Intel hardware intrinsic APIs #17637

Merged
merged 2 commits into from
Jun 18, 2018
Merged

Improve Intel hardware intrinsic APIs #17637

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
0157aab
Improve Intel hardware intrinsic APIs
FeiPengIntel Jun 14, 2018
5d8c06a
Simplify Avx.Extract non-const fallback
FeiPengIntel Jun 14, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 8 additions & 18 deletions src/System.Private.CoreLib/shared/System/Runtime/Intrinsics/X86/Avx.PlatformNotSupported.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -234,11 +234,6 @@ public static class Avx
/// </summary>
public static Vector256<float> DuplicateOddIndexed(Vector256<float> value) { throw new PlatformNotSupportedException(); }

/// <summary>
/// __int8 _mm256_extract_epi8 (__m256i a, const int index)
/// HELPER
/// </summary>
public static sbyte Extract(Vector256<sbyte> value, byte index) { throw new PlatformNotSupportedException(); }
/// <summary>
/// __int8 _mm256_extract_epi8 (__m256i a, const int index)
/// HELPER
Expand All @@ -248,11 +243,6 @@ public static class Avx
/// __int16 _mm256_extract_epi16 (__m256i a, const int index)
/// HELPER
/// </summary>
public static short Extract(Vector256<short> value, byte index) { throw new PlatformNotSupportedException(); }
/// <summary>
/// __int16 _mm256_extract_epi16 (__m256i a, const int index)
/// HELPER
/// </summary>
public static ushort Extract(Vector256<ushort> value, byte index) { throw new PlatformNotSupportedException(); }
/// <summary>
/// __int32 _mm256_extract_epi32 (__m256i a, const int index)
Expand Down Expand Up @@ -638,45 +628,45 @@ public static class Avx
/// __m128 _mm_maskload_ps (float const * mem_addr, __m128i mask)
/// VMASKMOVPS xmm, xmm, m128
/// </summary>
public static unsafe Vector128<float> MaskLoad(float* address, Vector128<uint> mask) { throw new PlatformNotSupportedException(); }
public static unsafe Vector128<float> MaskLoad(float* address, Vector128<float> mask) { throw new PlatformNotSupportedException(); }
/// <summary>
/// __m128d _mm_maskload_pd (double const * mem_addr, __m128i mask)
/// VMASKMOVPD xmm, xmm, m128
/// </summary>
public static unsafe Vector128<double> MaskLoad(double* address, Vector128<ulong> mask) { throw new PlatformNotSupportedException(); }
public static unsafe Vector128<double> MaskLoad(double* address, Vector128<double> mask) { throw new PlatformNotSupportedException(); }

/// <summary>
/// __m256 _mm256_maskload_ps (float const * mem_addr, __m256i mask)
/// VMASKMOVPS ymm, ymm, m256
/// </summary>
public static unsafe Vector256<float> MaskLoad(float* address, Vector256<uint> mask) { throw new PlatformNotSupportedException(); }
public static unsafe Vector256<float> MaskLoad(float* address, Vector256<float> mask) { throw new PlatformNotSupportedException(); }
/// <summary>
/// __m256d _mm256_maskload_pd (double const * mem_addr, __m256i mask)
/// VMASKMOVPD ymm, ymm, m256
/// </summary>
public static unsafe Vector256<double> MaskLoad(double* address, Vector256<ulong> mask) { throw new PlatformNotSupportedException(); }
public static unsafe Vector256<double> MaskLoad(double* address, Vector256<double> mask) { throw new PlatformNotSupportedException(); }

/// <summary>
/// void _mm_maskstore_ps (float * mem_addr, __m128i mask, __m128 a)
/// VMASKMOVPS m128, xmm, xmm
/// </summary>
public static unsafe void MaskStore(float* address, Vector128<float> mask, Vector128<uint> source) { throw new PlatformNotSupportedException(); }
public static unsafe void MaskStore(float* address, Vector128<float> mask, Vector128<float> source) { throw new PlatformNotSupportedException(); }
/// <summary>
/// void _mm_maskstore_pd (double * mem_addr, __m128i mask, __m128d a)
/// VMASKMOVPD m128, xmm, xmm
/// </summary>
public static unsafe void MaskStore(double* address, Vector128<double> mask, Vector128<ulong> source) { throw new PlatformNotSupportedException(); }
public static unsafe void MaskStore(double* address, Vector128<double> mask, Vector128<double> source) { throw new PlatformNotSupportedException(); }

/// <summary>
/// void _mm256_maskstore_ps (float * mem_addr, __m256i mask, __m256 a)
/// VMASKMOVPS m256, ymm, ymm
/// </summary>
public static unsafe void MaskStore(float* address, Vector256<float> mask, Vector256<uint> source) { throw new PlatformNotSupportedException(); }
public static unsafe void MaskStore(float* address, Vector256<float> mask, Vector256<float> source) { throw new PlatformNotSupportedException(); }
/// <summary>
/// void _mm256_maskstore_pd (double * mem_addr, __m256i mask, __m256d a)
/// VMASKMOVPD m256, ymm, ymm
/// </summary>
public static unsafe void MaskStore(double* address, Vector256<double> mask, Vector256<ulong> source) { throw new PlatformNotSupportedException(); }
public static unsafe void MaskStore(double* address, Vector256<double> mask, Vector256<double> source) { throw new PlatformNotSupportedException(); }

/// <summary>
/// __m256 _mm256_max_ps (__m256 a, __m256 b)
Expand Down
95 changes: 15 additions & 80 deletions src/System.Private.CoreLib/shared/System/Runtime/Intrinsics/X86/Avx.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@

using System;
using System.Runtime.Intrinsics;
using System.Runtime.CompilerServices;
using Internal.Runtime.CompilerServices;

namespace System.Runtime.Intrinsics.X86
{
Expand Down Expand Up @@ -235,64 +235,23 @@ public static class Avx
/// </summary>
public static Vector256<float> DuplicateOddIndexed(Vector256<float> value) => DuplicateOddIndexed(value);

/// <summary>
/// __int8 _mm256_extract_epi8 (__m256i a, const int index)
/// HELPER
/// </summary>
public static sbyte Extract(Vector256<sbyte> value, byte index)
{
unsafe
{
index &= 0x1F;
sbyte* buffer = stackalloc sbyte[32];
Store(buffer, value);
return buffer[index];
}
}

/// <summary>
/// __int8 _mm256_extract_epi8 (__m256i a, const int index)
/// HELPER
/// </summary>
public static byte Extract(Vector256<byte> value, byte index)
{
unsafe
{
index &= 0x1F;
byte* buffer = stackalloc byte[32];
Store(buffer, value);
return buffer[index];
}
return Unsafe.Add<byte>(ref Unsafe.As<Vector256<byte>, byte>(ref value), index & 0x1F);
Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Removed sbyte and short overloads of SSE2/SSE4.1/AVX Extract and simplify Avx.Extract non-const fallback as @jkotas's suggestion.

I would prepare the CoreFX counterpart if this PR looks good to you guys.

}

/// <summary>
/// __int16 _mm256_extract_epi16 (__m256i a, const int index)
/// HELPER
/// </summary>
public static short Extract(Vector256<short> value, byte index)
{
unsafe
{
index &= 0xF;
short* buffer = stackalloc short[16];
Store(buffer, value);
return buffer[index];
}
}

/// <summary>
/// __int16 _mm256_extract_epi16 (__m256i a, const int index)
/// HELPER
/// </summary>
public static ushort Extract(Vector256<ushort> value, byte index)
{
unsafe
{
index &= 0xF;
ushort* buffer = stackalloc ushort[16];
Store(buffer, value);
return buffer[index];
}
return Unsafe.Add<ushort>(ref Unsafe.As<Vector256<ushort>, ushort>(ref value), index & 0xF);
}

/// <summary>
Expand All @@ -301,13 +260,7 @@ public static ushort Extract(Vector256<ushort> value, byte index)
/// </summary>
public static int Extract(Vector256<int> value, byte index)
{
unsafe
{
index &= 0x7;
int* buffer = stackalloc int[8];
Store(buffer, value);
return buffer[index];
}
return Unsafe.Add<int>(ref Unsafe.As<Vector256<int>, int>(ref value), index & 0x7);
}

/// <summary>
Expand All @@ -316,13 +269,7 @@ public static int Extract(Vector256<int> value, byte index)
/// </summary>
public static uint Extract(Vector256<uint> value, byte index)
{
unsafe
{
index &= 0x7;
uint* buffer = stackalloc uint[8];
Store(buffer, value);
return buffer[index];
}
return Unsafe.Add<uint>(ref Unsafe.As<Vector256<uint>, uint>(ref value), index & 0x7);
}

/// <summary>
Expand All @@ -335,13 +282,7 @@ public static long Extract(Vector256<long> value, byte index)
{
throw new PlatformNotSupportedException();
}
unsafe
{
index &= 0x3;
long* buffer = stackalloc long[4];
Store(buffer, value);
return buffer[index];
}
return Unsafe.Add<long>(ref Unsafe.As<Vector256<long>, long>(ref value), index & 0x3);
}

/// <summary>
Expand All @@ -354,13 +295,7 @@ public static ulong Extract(Vector256<ulong> value, byte index)
{
throw new PlatformNotSupportedException();
}
unsafe
{
index &= 0x3;
ulong* buffer = stackalloc ulong[4];
Store(buffer, value);
return buffer[index];
}
return Unsafe.Add<ulong>(ref Unsafe.As<Vector256<ulong>, ulong>(ref value), index & 0x3);
}

/// <summary>
Expand Down Expand Up @@ -825,45 +760,45 @@ public static Vector256<T> InsertVector128<T>(Vector256<T> value, Vector128<T> d
/// __m128 _mm_maskload_ps (float const * mem_addr, __m128i mask)
/// VMASKMOVPS xmm, xmm, m128
/// </summary>
public static unsafe Vector128<float> MaskLoad(float* address, Vector128<uint> mask) => MaskLoad(address, mask);
public static unsafe Vector128<float> MaskLoad(float* address, Vector128<float> mask) => MaskLoad(address, mask);
/// <summary>
/// __m128d _mm_maskload_pd (double const * mem_addr, __m128i mask)
/// VMASKMOVPD xmm, xmm, m128
/// </summary>
public static unsafe Vector128<double> MaskLoad(double* address, Vector128<ulong> mask) => MaskLoad(address, mask);
public static unsafe Vector128<double> MaskLoad(double* address, Vector128<double> mask) => MaskLoad(address, mask);

/// <summary>
/// __m256 _mm256_maskload_ps (float const * mem_addr, __m256i mask)
/// VMASKMOVPS ymm, ymm, m256
/// </summary>
public static unsafe Vector256<float> MaskLoad(float* address, Vector256<uint> mask) => MaskLoad(address, mask);
public static unsafe Vector256<float> MaskLoad(float* address, Vector256<float> mask) => MaskLoad(address, mask);
/// <summary>
/// __m256d _mm256_maskload_pd (double const * mem_addr, __m256i mask)
/// VMASKMOVPD ymm, ymm, m256
/// </summary>
public static unsafe Vector256<double> MaskLoad(double* address, Vector256<ulong> mask) => MaskLoad(address, mask);
public static unsafe Vector256<double> MaskLoad(double* address, Vector256<double> mask) => MaskLoad(address, mask);

/// <summary>
/// void _mm_maskstore_ps (float * mem_addr, __m128i mask, __m128 a)
/// VMASKMOVPS m128, xmm, xmm
/// </summary>
public static unsafe void MaskStore(float* address, Vector128<float> mask, Vector128<uint> source) => MaskStore(address, mask, source);
public static unsafe void MaskStore(float* address, Vector128<float> mask, Vector128<float> source) => MaskStore(address, mask, source);
/// <summary>
/// void _mm_maskstore_pd (double * mem_addr, __m128i mask, __m128d a)
/// VMASKMOVPD m128, xmm, xmm
/// </summary>
public static unsafe void MaskStore(double* address, Vector128<double> mask, Vector128<ulong> source) => MaskStore(address, mask, source);
public static unsafe void MaskStore(double* address, Vector128<double> mask, Vector128<double> source) => MaskStore(address, mask, source);

/// <summary>
/// void _mm256_maskstore_ps (float * mem_addr, __m256i mask, __m256 a)
/// VMASKMOVPS m256, ymm, ymm
/// </summary>
public static unsafe void MaskStore(float* address, Vector256<float> mask, Vector256<uint> source) => MaskStore(address, mask, source);
public static unsafe void MaskStore(float* address, Vector256<float> mask, Vector256<float> source) => MaskStore(address, mask, source);
/// <summary>
/// void _mm256_maskstore_pd (double * mem_addr, __m256i mask, __m256d a)
/// VMASKMOVPD m256, ymm, ymm
/// </summary>
public static unsafe void MaskStore(double* address, Vector256<double> mask, Vector256<ulong> source) => MaskStore(address, mask, source);
public static unsafe void MaskStore(double* address, Vector256<double> mask, Vector256<double> source) => MaskStore(address, mask, source);

/// <summary>
/// __m256 _mm256_max_ps (__m256 a, __m256 b)
Expand Down
5 changes: 0 additions & 5 deletions src/System.Private.CoreLib/shared/System/Runtime/Intrinsics/X86/Sse2.PlatformNotSupported.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -609,11 +609,6 @@ public static class Sse2
/// </summary>
public static Vector128<double> DivideScalar(Vector128<double> left, Vector128<double> right) { throw new PlatformNotSupportedException(); }

/// <summary>
/// int _mm_extract_epi16 (__m128i a, int immediate)
/// PEXTRW reg, xmm, imm8
/// </summary>
public static short Extract(Vector128<short> value, byte index) { throw new PlatformNotSupportedException(); }
/// <summary>
/// int _mm_extract_epi16 (__m128i a, int immediate)
/// PEXTRW reg, xmm, imm8
Expand Down
5 changes: 0 additions & 5 deletions src/System.Private.CoreLib/shared/System/Runtime/Intrinsics/X86/Sse2.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -610,11 +610,6 @@ public static class Sse2
/// </summary>
public static Vector128<double> DivideScalar(Vector128<double> left, Vector128<double> right) => DivideScalar(left, right);

/// <summary>
/// int _mm_extract_epi16 (__m128i a, int immediate)
/// PEXTRW reg, xmm, imm8
/// </summary>
public static short Extract(Vector128<short> value, byte index) => Extract(value, index);
/// <summary>
/// int _mm_extract_epi16 (__m128i a, int immediate)
/// PEXTRW reg, xmm, imm8
Expand Down
7 changes: 1 addition & 6 deletions ...System.Private.CoreLib/shared/System/Runtime/Intrinsics/X86/Sse41.PlatformNotSupported.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -178,11 +178,6 @@ public static class Sse41
/// </summary>
public static Vector128<double> DotProduct(Vector128<double> left, Vector128<double> right, byte control) { throw new PlatformNotSupportedException(); }

/// <summary>
/// int _mm_extract_epi8 (__m128i a, const int imm8)
/// PEXTRB reg/m8, xmm, imm8
/// </summary>
public static sbyte Extract(Vector128<sbyte> value, byte index) { throw new PlatformNotSupportedException(); }
/// <summary>
/// int _mm_extract_epi8 (__m128i a, const int imm8)
/// PEXTRB reg/m8, xmm, imm8
Expand Down Expand Up @@ -283,7 +278,7 @@ public static class Sse41
/// __m128 _mm_insert_ps (__m128 a, __m128 b, const int imm8)
/// INSERTPS xmm, xmm/m32, imm8
/// </summary>
public static Vector128<float> Insert(Vector128<float> value, float data, byte index) { throw new PlatformNotSupportedException(); }
public static Vector128<float> Insert(Vector128<float> value, Vector128<float> data, byte index) { throw new PlatformNotSupportedException(); }

/// <summary>
/// __m128i _mm_max_epi8 (__m128i a, __m128i b)
Expand Down
7 changes: 1 addition & 6 deletions src/System.Private.CoreLib/shared/System/Runtime/Intrinsics/X86/Sse41.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -178,11 +178,6 @@ public static class Sse41
/// </summary>
public static Vector128<double> DotProduct(Vector128<double> left, Vector128<double> right, byte control) => DotProduct(left, right, control);

/// <summary>
/// int _mm_extract_epi8 (__m128i a, const int imm8)
/// PEXTRB reg/m8, xmm, imm8
/// </summary>
public static sbyte Extract(Vector128<sbyte> value, byte index) => Extract(value, index);
/// <summary>
/// int _mm_extract_epi8 (__m128i a, const int imm8)
/// PEXTRB reg/m8, xmm, imm8
Expand Down Expand Up @@ -283,7 +278,7 @@ public static class Sse41
/// __m128 _mm_insert_ps (__m128 a, __m128 b, const int imm8)
/// INSERTPS xmm, xmm/m32, imm8
/// </summary>
public static Vector128<float> Insert(Vector128<float> value, float data, byte index) => Insert(value, data, index);
public static Vector128<float> Insert(Vector128<float> value, Vector128<float> data, byte index) => Insert(value, data, index);

/// <summary>
/// __m128i _mm_max_epi8 (__m128i a, __m128i b)
Expand Down
6 changes: 0 additions & 6 deletions tests/src/JIT/HardwareIntrinsics/X86/Avx/Avx_r.csproj
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,12 +45,8 @@
<Compile Include="DuplicateEvenIndexed.Single.cs" />
<Compile Include="DuplicateOddIndexed.Single.cs" />
<Compile Include="Extract.Byte.1.cs" />
<Compile Include="Extract.SByte.1.cs" />
<Compile Include="Extract.Byte.20.cs" />
<Compile Include="Extract.SByte.20.cs" />
<Compile Include="Extract.Int16.1.cs" />
<Compile Include="Extract.UInt16.1.cs" />
<Compile Include="Extract.Int16.11.cs" />
<Compile Include="Extract.UInt16.11.cs" />
<Compile Include="Extract.Int32.1.cs" />
<Compile Include="Extract.UInt32.1.cs" />
Expand All @@ -61,8 +57,6 @@
<Compile Include="Extract.Int64.3.cs" />
<Compile Include="Extract.UInt64.3.cs" />
<Compile Include="Extract.Byte.52.cs" />
<Compile Include="Extract.SByte.52.cs" />
<Compile Include="Extract.Int16.27.cs" />
<Compile Include="Extract.UInt16.27.cs" />
<Compile Include="Extract.Int32.22.cs" />
<Compile Include="Extract.UInt32.22.cs" />
Expand Down
6 changes: 0 additions & 6 deletions tests/src/JIT/HardwareIntrinsics/X86/Avx/Avx_ro.csproj
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,12 +45,8 @@
<Compile Include="DuplicateEvenIndexed.Single.cs" />
<Compile Include="DuplicateOddIndexed.Single.cs" />
<Compile Include="Extract.Byte.1.cs" />
<Compile Include="Extract.SByte.1.cs" />
<Compile Include="Extract.Byte.20.cs" />
<Compile Include="Extract.SByte.20.cs" />
<Compile Include="Extract.Int16.1.cs" />
<Compile Include="Extract.UInt16.1.cs" />
<Compile Include="Extract.Int16.11.cs" />
<Compile Include="Extract.UInt16.11.cs" />
<Compile Include="Extract.Int32.1.cs" />
<Compile Include="Extract.UInt32.1.cs" />
Expand All @@ -61,8 +57,6 @@
<Compile Include="Extract.Int64.3.cs" />
<Compile Include="Extract.UInt64.3.cs" />
<Compile Include="Extract.Byte.52.cs" />
<Compile Include="Extract.SByte.52.cs" />
<Compile Include="Extract.Int16.27.cs" />
<Compile Include="Extract.UInt16.27.cs" />
<Compile Include="Extract.Int32.22.cs" />
<Compile Include="Extract.UInt32.22.cs" />
Expand Down
Loading