projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a0372de) | patch
[X86][InstCombine] Teach SimplifyDemandedVectorElts to handle scalar round intrinsics...
authorCraig Topper <craig.topper@gmail.com>
Wed, 14 Dec 2016 03:17:30 +0000 (03:17 +0000)
committerCraig Topper <craig.topper@gmail.com>
Wed, 14 Dec 2016 03:17:30 +0000 (03:17 +0000)
commiteb6a20e79ed4a5ac8d5f61f7372bb633c7362378
treef582a662bab3c3b210043da9047358b16053f51ctree | snapshot
parenta0372dec261d137973b29ed14376230c372b4290commit | diff
[X86][InstCombine] Teach SimplifyDemandedVectorElts to handle scalar round intrinsics more correctly.

Now we only pass bit 0 of the DemandedElts to optimize operand 1 as we recurse since the upper bits are unused. Similarly we clear bit 0 for optimizing operand 0.

Also calculate UndefElts correctly.

Simplify InstCombineCalls for these instrinics to just call SimplifyDemandedVectorElts for the call instrution to reuse this support.

llvm-svn: 289629
llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp diff | blob | history
llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp diff | blob | history
Domain: System / Toolchain;