Shuxin Yang b1ccfb3a54 [Fast-math] Disable "(C1/X)*C2 => (C1*C2)/X" if C1/X has multiple uses. ... If "C1/X" were having multiple uses, the only benefit of this transformation is to potentially shorten critical path. But it is at the cost of instroducing additional div. The additional div may or may not incur cost depending on how div is implemented. If it is implemented using Newton–Raphson iteration, it dosen't seem to incur any cost (FIXME). However, if the div blocks the entire pipeline, that sounds to be pretty expensive. Let CodeGen to take care this transformation. This patch sees 6% on a benchmark. rdar://15032743 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@191037 91177308-0d34-0410-b5e6-96231b3b80d8		2013-09-19 21:13:46 +00:00
..
Analysis	Fix a constant folding address space place I missed.	2013-09-17 23:23:16 +00:00
AsmParser	Implement function prefix data as an IR feature.	2013-09-16 01:08:15 +00:00
Bitcode	Implement function prefix data as an IR feature.	2013-09-16 01:08:15 +00:00
CodeGen	Unshift the GDB index/GNU pubnames constants modified in r191025	2013-09-19 20:40:26 +00:00
DebugInfo
ExecutionEngine	llvm-c: Add LLVMGetPointerToFunction	2013-09-19 19:55:06 +00:00
IR	Revert "Encapsulate PassManager debug flags to avoid static init and cxa_exit."	2013-09-19 06:02:43 +00:00
IRReader
Linker	Implement function prefix data as an IR feature.	2013-09-16 01:08:15 +00:00
MC	Remove extraneous space, the asm printing infrastructure adds a space	2013-09-19 18:41:40 +00:00
Object	ELF: Add support for the exclude section bit for gas compat.	2013-09-15 19:53:20 +00:00
Option	Fix another mistake in r190442.	2013-09-10 23:22:56 +00:00
Support	Unshift the GDB index/GNU pubnames constants modified in r191025	2013-09-19 20:40:26 +00:00
TableGen
Target	X86 horizontal vector reduction cost model	2013-09-19 17:48:48 +00:00
Transforms	[Fast-math] Disable "(C1/X)*C2 => (C1*C2)/X" if C1/X has multiple uses.	2013-09-19 21:13:46 +00:00
CMakeLists.txt
LLVMBuild.txt
Makefile