projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 4fa12ac) | patch
[X86] Use sse_load_f32/f64 and timm in patterns for memory form of vgetmantss/sd.
authorCraig Topper <craig.topper@intel.com>
Sat, 21 Sep 2019 06:44:29 +0000 (06:44 +0000)
committerCraig Topper <craig.topper@intel.com>
Sat, 21 Sep 2019 06:44:29 +0000 (06:44 +0000)
commit04682939eb7e5ff06ec7e50cbe205b5c0069285f
tree9cbd34fc1c1ad7667679ec83f9f7210a1502315ftree | snapshot
parent4fa12ac92ced6dafdaf1e7c7e7a4b86d774e0b83commit | diff
[X86] Use sse_load_f32/f64 and timm in patterns for memory form of vgetmantss/sd.

Previously we only matched scalar_to_vector and scalar load, but
we should be able to narrow a vector load or match vzload.

Also need to match TargetConstant instead of Constant. The register
patterns were previously updated, but not the memory patterns.

llvm-svn: 372458
llvm/lib/Target/X86/X86InstrAVX512.td diff | blob | history
llvm/test/CodeGen/X86/avx512-intrinsics.ll diff | blob | history
Domain: System / Toolchain;