diff options
| author | Craig Topper <craig.topper@intel.com> | 2017-12-15 20:57:18 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2017-12-15 20:57:18 +0000 |
| commit | f08ab74ae3a2e64b475d67ec26413b3712f438ac (patch) | |
| tree | 7cc1667d0968209e380b0eee1c2ecc6bdc5e3eb1 | |
| parent | cf29eb8c22f66f03d460c9a6713d15b626a1bc81 (diff) | |
| download | bcm5719-llvm-f08ab74ae3a2e64b475d67ec26413b3712f438ac.tar.gz bcm5719-llvm-f08ab74ae3a2e64b475d67ec26413b3712f438ac.zip | |
[X86] Remove unnecessary TODO.
When I wrote it I thought we were missing a potential optimization for KNL. But investigating further shows that for KNL we still do the optimal thing by widening to v4f32 and then using special isel patterns to widen again to zmm a register.
llvm-svn: 320862
| -rw-r--r-- | llvm/lib/Target/X86/X86ISelLowering.cpp | 1 |
1 files changed, 0 insertions, 1 deletions
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp index 8468e4ed013..0429f65e73d 100644 --- a/llvm/lib/Target/X86/X86ISelLowering.cpp +++ b/llvm/lib/Target/X86/X86ISelLowering.cpp @@ -1566,7 +1566,6 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM, setTruncStoreAction(MVT::v4i32, MVT::v4i16, Legal); if (Subtarget.hasDQI()) { - // TODO: these shouldn't require VLX. We can widen to 512-bit with AVX512F. // Fast v2f32 SINT_TO_FP( v2i64 ) custom conversion. // v2f32 UINT_TO_FP is already custom under SSE2. setOperationAction(ISD::SINT_TO_FP, MVT::v2f32, Custom); |
