Skip to content

Optimize WithLower, WithUpper, Create, AsInt64, AsUInt64, AsDouble with ARM64 hardware intrinsics #37139

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

Merged
merged 12 commits into from
Jun 3, 2020
Merged

Optimize WithLower, WithUpper, Create, AsInt64, AsUInt64, AsDouble with ARM64 hardware intrinsics #37139

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
2c78526
Optimize Vector64.AsDouble(), Vector64.AsInt64(), Vector64.AsUInt64()
kunalspathak May 28, 2020
5e856ba
Optimize Vector128.WithUpper()
kunalspathak May 28, 2020
7ee86cf
Inline GetElement() as paramter
kunalspathak May 28, 2020
957f4cc
Combine Insert/GetElement in JIT
kunalspathak May 28, 2020
a78219d
Optimize Vector128.WithLower()
kunalspathak May 28, 2020
b5761f0
Optimize Vector128.Create(Vector64, Vector64)
kunalspathak May 28, 2020
48f09b7
minor fix
kunalspathak May 29, 2020
7d582db
review comments
kunalspathak Jun 1, 2020
e6b0e8b
Change the ifdef from ARMARCH to ARM64
kunalspathak Jun 1, 2020
fddba3e
Mark optimized methods to be AggresiveInlining
kunalspathak Jun 1, 2020
c480f42
Revert optimization in JIT
kunalspathak Jun 1, 2020
b5dcb16
Use ToVector128Unsafe()
kunalspathak Jun 1, 2020
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
36 changes: 19 additions & 17 deletions src/coreclr/src/jit/hwintrinsic.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -646,24 +646,26 @@ static bool isSupportedBaseType(NamedIntrinsic intrinsic, var_types baseType)
(intrinsic == NI_Vector256_get_Zero) || (intrinsic == NI_Vector256_GetElement) ||
(intrinsic == NI_Vector256_WithElement) || (intrinsic == NI_Vector256_GetLower) ||
(intrinsic == NI_Vector256_ToScalar));
#else
assert((intrinsic == NI_Vector64_AsByte) || (intrinsic == NI_Vector64_AsInt16) ||
(intrinsic == NI_Vector64_AsInt32) || (intrinsic == NI_Vector64_AsSByte) ||
#endif // TARGET_XARCH
#ifdef TARGET_ARM64
assert((intrinsic == NI_Vector64_AsByte) || (intrinsic == NI_Vector64_AsDouble) ||
(intrinsic == NI_Vector64_AsInt16) || (intrinsic == NI_Vector64_AsInt32) ||
(intrinsic == NI_Vector64_AsInt64) || (intrinsic == NI_Vector64_AsSByte) ||
(intrinsic == NI_Vector64_AsSingle) || (intrinsic == NI_Vector64_AsUInt16) ||
(intrinsic == NI_Vector64_AsUInt32) || (intrinsic == NI_Vector64_get_AllBitsSet) ||
(intrinsic == NI_Vector64_get_Count) || (intrinsic == NI_Vector64_get_Zero) ||
(intrinsic == NI_Vector64_GetElement) || (intrinsic == NI_Vector64_ToScalar) ||
(intrinsic == NI_Vector64_ToVector128) || (intrinsic == NI_Vector64_ToVector128Unsafe) ||
(intrinsic == NI_Vector128_As) || (intrinsic == NI_Vector128_AsByte) ||
(intrinsic == NI_Vector128_AsDouble) || (intrinsic == NI_Vector128_AsInt16) ||
(intrinsic == NI_Vector128_AsInt32) || (intrinsic == NI_Vector128_AsInt64) ||
(intrinsic == NI_Vector128_AsSByte) || (intrinsic == NI_Vector128_AsSingle) ||
(intrinsic == NI_Vector128_AsUInt16) || (intrinsic == NI_Vector128_AsUInt32) ||
(intrinsic == NI_Vector128_AsUInt64) || (intrinsic == NI_Vector128_get_AllBitsSet) ||
(intrinsic == NI_Vector128_get_Count) || (intrinsic == NI_Vector128_get_Zero) ||
(intrinsic == NI_Vector128_GetElement) || (intrinsic == NI_Vector128_GetLower) ||
(intrinsic == NI_Vector128_ToScalar));
#endif
(intrinsic == NI_Vector64_AsUInt32) || (intrinsic == NI_Vector64_AsUInt64) ||
(intrinsic == NI_Vector64_get_AllBitsSet) || (intrinsic == NI_Vector64_get_Count) ||
(intrinsic == NI_Vector64_get_Zero) || (intrinsic == NI_Vector64_GetElement) ||
(intrinsic == NI_Vector64_ToScalar) || (intrinsic == NI_Vector64_ToVector128) ||
(intrinsic == NI_Vector64_ToVector128Unsafe) || (intrinsic == NI_Vector128_As) ||
(intrinsic == NI_Vector128_AsByte) || (intrinsic == NI_Vector128_AsDouble) ||
(intrinsic == NI_Vector128_AsInt16) || (intrinsic == NI_Vector128_AsInt32) ||
(intrinsic == NI_Vector128_AsInt64) || (intrinsic == NI_Vector128_AsSByte) ||
(intrinsic == NI_Vector128_AsSingle) || (intrinsic == NI_Vector128_AsUInt16) ||
(intrinsic == NI_Vector128_AsUInt32) || (intrinsic == NI_Vector128_AsUInt64) ||
(intrinsic == NI_Vector128_get_AllBitsSet) || (intrinsic == NI_Vector128_get_Count) ||
(intrinsic == NI_Vector128_get_Zero) || (intrinsic == NI_Vector128_GetElement) ||
(intrinsic == NI_Vector128_GetLower) || (intrinsic == NI_Vector128_ToScalar));
#endif // TARGET_ARM64
return false;
}

Expand Down
3 changes: 3 additions & 0 deletions src/coreclr/src/jit/hwintrinsicarm64.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -359,12 +359,15 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic intrinsic,
switch (intrinsic)
{
case NI_Vector64_AsByte:
case NI_Vector64_AsDouble:
case NI_Vector64_AsInt16:
case NI_Vector64_AsInt32:
case NI_Vector64_AsInt64:
case NI_Vector64_AsSByte:
case NI_Vector64_AsSingle:
case NI_Vector64_AsUInt16:
case NI_Vector64_AsUInt32:
case NI_Vector64_AsUInt64:
case NI_Vector128_As:
case NI_Vector128_AsByte:
case NI_Vector128_AsDouble:
Expand Down
3 changes: 3 additions & 0 deletions src/coreclr/src/jit/hwintrinsiclistarm64.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,12 +17,15 @@
// ***************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************************
// Vector64 Intrinsics
HARDWARE_INTRINSIC(Vector64, AsByte, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsDouble, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsInt16, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsInt32, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsInt64, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsSByte, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsSingle, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsUInt16, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsUInt32, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, AsUInt64, 8, 1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
HARDWARE_INTRINSIC(Vector64, Create, 8, -1, {INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_invalid, INS_mov, INS_mov, INS_invalid, INS_invalid}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
HARDWARE_INTRINSIC(Vector64, CreateScalarUnsafe, 8, 1, {INS_ins, INS_ins, INS_ins, INS_ins, INS_ins, INS_ins, INS_invalid, INS_invalid, INS_fmov, INS_invalid}, HW_Category_SimpleSIMD, HW_Flag_SupportsContainment|HW_Flag_SpecialCodeGen)
HARDWARE_INTRINSIC(Vector64, get_AllBitsSet, 8, 0, {INS_mvni, INS_mvni, INS_mvni, INS_mvni, INS_mvni, INS_mvni, INS_mvni, INS_mvni, INS_mvni, INS_mvni}, HW_Category_Helper, HW_Flag_SpecialImport|HW_Flag_SpecialCodeGen)
Expand Down
Loading