projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: f27a40a) | patch
Cache basic block bounds instead of asking SlotIndexes::getMBBRange all the time.
authorJakob Stoklund Olesen <stoklund@2pi.dk>
Thu, 3 Mar 2011 03:41:29 +0000 (03:41 +0000)
committerJakob Stoklund Olesen <stoklund@2pi.dk>
Thu, 3 Mar 2011 03:41:29 +0000 (03:41 +0000)
commit36d61863bc83bd2301e0224adc560098b35ec0dc
treef263712c35c5e76d30493c5b80e9bfe9175d2099tree | snapshot
parentf27a40a9717c019fd07990483fb475b544bc895ecommit | diff
Cache basic block bounds instead of asking SlotIndexes::getMBBRange all the time.

This speeds up the greedy register allocator by 15%.
DenseMap is not as fast as one might hope.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@126921 91177308-0d34-0410-b5e6-96231b3b80d8
lib/CodeGen/RegAllocGreedy.cpp diff | blob | history
lib/CodeGen/SplitKit.cpp diff | blob | history
lib/CodeGen/SplitKit.h diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors