projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 2d53a37) | patch
ARMTTI: We don't have 16 allocatable scalar registers
authorArnold Schwaighofer <aschwaighofer@apple.com>
Sat, 1 Feb 2014 18:00:25 +0000 (18:00 +0000)
committerArnold Schwaighofer <aschwaighofer@apple.com>
Sat, 1 Feb 2014 18:00:25 +0000 (18:00 +0000)
commit991dd3bb9284c49549835687285c8bb4af81e2e7
treef5644b3d1b18e7f23ee6c157d48d7e9b2f1ee5e2tree | snapshot
parent2d53a375732b044a3ec44bd41f17feed0eb8f064commit | diff
ARMTTI: We don't have 16 allocatable scalar registers

This caused an regression on libquantum after enabling the new loop vectorizer
unroll heuristics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200616 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/ARM/ARMTargetTransformInfo.cpp diff | blob | history
test/Transforms/LoopVectorize/ARM/arm-unroll.ll diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors