Benjamin Kramer 0463564612 CodeGenPrepare: Add a transform to turn selects into branches in some cases. ... This came up when a change in block placement formed a cmov and slowed down a hot loop by 50%: ucomisd (%rdi), %xmm0 cmovbel %edx, %esi cmov is a really bad choice in this context because it doesn't get branch prediction. If we emit it as a branch, an out-of-order CPU can do a better job (if the branch is predicted right) and avoid waiting for the slow load+compare instruction to finish. Of course it won't help if the branch is unpredictable, but those are really rare in practice. This patch uses a dumb conservative heuristic, it turns all cmovs that have one use and a direct memory operand into branches. cmovs usually save some code size, so we disable the transform in -Os mode. In-Order architectures are unlikely to benefit as well, those are included in the "predictableSelectIsExpensive" flag. It would be better to reuse branch probability info here, but BPI doesn't support select instructions currently. It would make sense to use the same heuristics as the if-converter pass, which does the opposite direction of this transform. Test suite shows a small improvement here and there on corei7-level machines, but the actual results depend a lot on the used microarchitecture. The transformation is currently disabled by default and available by passing the -enable-cgp-select2branch flag to the code generator. Thanks to Chandler for the initial test case to him and Evan Cheng for providing me with comments and test-suite numbers that were more stable than mine :) llvm-svn: 156234		2012-05-05 12:49:22 +00:00
..
Analysis	FileCheck-ize tests.	2012-04-24 10:45:44 +00:00
Archive
Assembler	Remove old 'grep' lines.	2012-04-08 11:53:54 +00:00
Bindings/Ocaml
Bitcode	FileCheckize these testcases.	2012-04-08 11:00:38 +00:00
BugPoint
CodeGen	CodeGenPrepare: Add a transform to turn selects into branches in some cases.	2012-05-05 12:49:22 +00:00
DebugInfo	X86: Use StackRegister instead of FrameRegister in getFrameIndexReference (to generate debug info for local variables) if stack needs realignment	2012-05-01 15:16:06 +00:00
ExecutionEngine	This patch improves the MCJIT runtime dynamic loader by adding new handling	2012-04-12 20:13:57 +00:00
Feature
Instrumentation	[tsan] Atomic support for ThreadSanitizer, patch by Dmitry Vyukov	2012-04-27 07:31:53 +00:00
Integer
Linker	Remove llvm-ld and llvm-stub (which is only used by llvm-ld).	2012-04-19 19:27:54 +00:00
MC	Fix issues with the ARM bl and blx thumb instructions and the J1 and J2 bits	2012-05-03 22:41:56 +00:00
Object
Other	Reapply r155682, making constant folding more consistent, with a fix to work	2012-04-27 17:50:22 +00:00
Scripts
TableGen
Transforms	Small fix in InstCombineCasts.cpp. Restored "alloca + bitcast" reducing for case when alloca's size is calculated within the "add/sub/... nsw".	2012-05-05 07:09:40 +00:00
Unit
Verifier	Remove support for the special 'fast' value for fpmath accuracy for the moment.	2012-04-16 19:39:33 +00:00
YAMLParser	Add YAML parser to Support.	2012-04-03 23:09:22 +00:00
CMakeLists.txt	Removes json-bench from the test dependencies.	2012-04-20 13:45:49 +00:00
lit.cfg	Remove llvm-ld and llvm-stub (which is only used by llvm-ld).	2012-04-19 19:27:54 +00:00
lit.site.cfg.in	Revert r153694. It was causing failures in the buildbots.	2012-03-29 23:23:59 +00:00
Makefile	Revert r153694. It was causing failures in the buildbots.	2012-03-29 23:23:59 +00:00
Makefile.tests
site.exp.in
TestRunner.sh