projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 181826c) | patch
[X86] AVX512: Simplify logic in isCDisp8
authorAdam Nemet <anemet@apple.com>
Fri, 11 Jul 2014 05:23:12 +0000 (05:23 +0000)
committerAdam Nemet <anemet@apple.com>
Fri, 11 Jul 2014 05:23:12 +0000 (05:23 +0000)
commitfcc56eb31fb98c1a6b91a545f9cf7652d9291bd8
tree6622c0041cba724fcd4fcb7e6928edb6cd634746tree | snapshot
parent181826c6233c268e1fab14ca3f6cf6ca99059d93commit | diff
[X86] AVX512: Simplify logic in isCDisp8

It was computing the VL/n case as:
  MemObjSize = VectorByteSize / ElemByteSize / Divider * ElemByteSize

ElemByteSize not only falls out but VectorByteSize/Divider now actually
matches the definition of VL/n.

Also some formatting fixes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212794 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors