bdd062dae0f27953ec77039a171d0f2b8fe72dfc - toolchain/llvm-project

commit	bdd062dae0f27953ec77039a171d0f2b8fe72dfc	[log] [tgz]
author	Evan Cheng <evan.cheng@apple.com>	Thu May 20 06:13:19 2010 +0000
committer	Evan Cheng <evan.cheng@apple.com>	Thu May 20 06:13:19 2010 +0000
tree	4622fd6ccaeacd89ac38f69a12a317d31cf3f31e
parent	aa2ac80eaaaea7d99e1efecf52abb4b167dc11c8 [diff]

Add a hybrid bottom up scheduler that reduce register usage while avoiding
pipeline stall. It's useful for targets like ARM cortex-a8. NEON has a lot
of long latency instructions so a strict register pressure reduction
scheduler does not work well.
Early experiments show this speeds up some NEON loops by over 30%.

llvm-svn: 104216

llvm/include/llvm/CodeGen/LinkAllCodegenComponents.h[diff]
llvm/include/llvm/CodeGen/ScheduleDAG.h[diff]
llvm/include/llvm/CodeGen/SchedulerRegistry.h[diff]
llvm/include/llvm/Target/TargetMachine.h[diff]
llvm/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp[diff]
llvm/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp[diff]
llvm/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.h[diff]
llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp[diff]

8 files changed

tree: 4622fd6ccaeacd89ac38f69a12a317d31cf3f31e

clang/
compiler-rt/
libcxx/
llvm/