diff options
| author | Craig Topper <craig.topper@intel.com> | 2018-05-28 19:33:11 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2018-05-28 19:33:11 +0000 |
| commit | dcfcfdb0d166fff8388bdd2edc5a2948054c9da1 (patch) | |
| tree | ed53f4e61304ba4f384eb0cc2ee6716e91a4d921 /llvm/lib/Target/X86/X86ISelLowering.h | |
| parent | 6b545182fbab96e6656a09e6d75d66935843d356 (diff) | |
| download | bcm5719-llvm-dcfcfdb0d166fff8388bdd2edc5a2948054c9da1.tar.gz bcm5719-llvm-dcfcfdb0d166fff8388bdd2edc5a2948054c9da1.zip | |
[X86] Converge X86ISD::VPERMV3 and X86ISD::VPERMIV3 to a single opcode.
These do the same thing with the first and second sources swapped. They previously came from separate intrinsics that specified different masking behavior. But we can cover that with isel patterns and a single node.
This is a step towards reducing the number of intrinsics needed.
A bunch of tests change because we are now biased to choosing VPERMT over VPERMI when there is nothing to signal that commuting is beneficial.
llvm-svn: 333383
Diffstat (limited to 'llvm/lib/Target/X86/X86ISelLowering.h')
| -rw-r--r-- | llvm/lib/Target/X86/X86ISelLowering.h | 4 |
1 files changed, 0 insertions, 4 deletions
diff --git a/llvm/lib/Target/X86/X86ISelLowering.h b/llvm/lib/Target/X86/X86ISelLowering.h index 58d0c9d92af..dd6be5b8e48 100644 --- a/llvm/lib/Target/X86/X86ISelLowering.h +++ b/llvm/lib/Target/X86/X86ISelLowering.h @@ -428,10 +428,6 @@ namespace llvm { // Res = VPERMV3 V0, MaskV, V1 VPERMV3, - // 3-op Variable Permute overwriting the index (VPERMI2). - // Res = VPERMIV3 V0, MaskV, V1 - VPERMIV3, - // Bitwise ternary logic. VPTERNLOG, // Fix Up Special Packed Float32/64 values. |
