OSDN Git Service

[CostModel][X86] Match 256-bit vector shift 'splat' costs for AVX2 and above
authorSimon Pilgrim <llvm-dev@redking.me.uk>
Sat, 7 Jan 2017 21:47:10 +0000 (21:47 +0000)
committerSimon Pilgrim <llvm-dev@redking.me.uk>
Sat, 7 Jan 2017 21:47:10 +0000 (21:47 +0000)
commit371d289738058e576541fafc7b8b242e8b6f0376
tree885d6806840e74e0494e99a7d3439effb7c02f5e
parente1ddc8e7d2668302343946a6bf70f011a446942e
[CostModel][X86] Match 256-bit vector shift 'splat' costs for AVX2 and above

We were matching against general vector shift costs before the uniform splat costs

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291365 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/X86TargetTransformInfo.cpp
test/Analysis/CostModel/X86/vshift-ashr-cost.ll
test/Analysis/CostModel/X86/vshift-lshr-cost.ll
test/Analysis/CostModel/X86/vshift-shl-cost.ll