Benjamin Kramer 1af132dcf3 LoopVectorizer: Emit memory checks into their own basic block. ... This separates the check for "too few elements to run the vector loop" from the "memory overlap" check, giving a lot nicer code and allowing to skip the memory checks when we're not going to execute the vector code anyways. We still leave the decision of whether to emit the memory checks as branches or setccs, but it seems to be doing a good job. If ugly code pops up we may want to emit them as separate blocks too. Small speedup on MultiSource/Benchmarks/MallocBench/espresso. Most of this is legwork to allow multiple bypass blocks while updating PHIs, dominators and loop info. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172902 91177308-0d34-0410-b5e6-96231b3b80d8		2013-01-19 13:57:58 +00:00
..
ADCE
ArgumentPromotion	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 02:33:22 +00:00
BBVectorize	Switch BBVectorize to directly depend on having a TTI analysis.	2013-01-07 10:22:36 +00:00
BlockPlacement
BranchFolding
CodeExtractor
CodeGenPrepare	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 01:28:40 +00:00
ConstantMerge	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 01:28:40 +00:00
ConstProp	llvm/ConstantFolding.cpp: Make ReadDataFromGlobal() and FoldReinterpretLoadFromConstPtr() Big-endian-aware.	2012-11-08 20:34:25 +00:00
CorrelatedValuePropagation	CorrelatedPropagation: BasicBlock::removePredecessor can simplify PHI nodes. If the it's the condition of a SwitchInst, reload it.	2012-09-28 10:42:50 +00:00
DeadArgElim	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
DeadStoreElimination	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
EarlyCSE	There was a switch fall-through in the parser for textual LLVM that caused	2013-01-07 13:32:38 +00:00
FunctionAttrs
GlobalDCE
GlobalOpt	Fix TryToShrinkGlobalToBoolean in GlobalOpt, so that it does not discard address spaces.	2013-01-10 10:31:11 +00:00
GVN	Memory Dependence Analysis: fix a miscompile that uses DT to approxmiate the	2013-01-04 19:19:47 +00:00
IndVarSimplify	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 02:33:22 +00:00
Inline	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 14:04:36 +00:00
InstCombine	Reverting r171325 & r172363. This was causing a mis-compile on the self-hosted LTO build bots.	2013-01-17 21:28:46 +00:00
InstSimplify	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 02:33:22 +00:00
Internalize	Change the internalize pass to internalize all symbols when given an empty	2012-10-26 18:47:48 +00:00
IPConstantProp
JumpThreading	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
LCSSA
LICM	LICM: Hoist insertvalue/extractvalue out of loops.	2013-01-09 18:12:03 +00:00
LoopDeletion	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
LoopIdiom	PR14904: Segmentation fault running pass 'Recognize loop idioms'	2013-01-10 23:32:01 +00:00
LoopRotate	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 02:33:22 +00:00
LoopSimplify
LoopStrengthReduce	Remove triple from this test, it makes it fail when X86 TTI is missing.	2013-01-16 13:19:59 +00:00
LoopUnroll	Add a new attribute, 'noduplicate'. If a function contains a noduplicate call, the call cannot be duplicated - Jump threading, loop unrolling, loop unswitching, and loop rotation are inhibited if they would duplicate the call.	2012-12-20 16:04:27 +00:00
LoopUnswitch	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 02:33:22 +00:00
LoopVectorize	LoopVectorizer: Emit memory checks into their own basic block.	2013-01-19 13:57:58 +00:00
LowerAtomic
LowerExpectIntrinsic
LowerInvoke
LowerSwitch
Mem2Reg
MemCpyOpt	In my recent change to avoid use of underaligned memory I didn't notice that	2012-10-04 13:53:21 +00:00
MergeFunc	CastInst::castIsValid should return true if the dest type is the same as	2013-01-10 23:22:53 +00:00
MetaRenamer	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
ObjCARC	Added test for r172599 which fixes bugzilla://14584,rdar://11744105.	2013-01-16 21:07:18 +00:00
PhaseOrdering	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 01:28:40 +00:00
PruneEH
Reassociate	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 02:33:22 +00:00
Reg2Mem	Make sure we don't emit instructions before a landingpad instruction.	2013-01-08 10:51:32 +00:00
ScalarRepl	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
SCCP	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
SimplifyCFG	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
SimplifyLibCalls	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2012-12-30 01:28:40 +00:00
Sink
SROA	Rename the test so that we can add additional vectors-of-pointers tests	2012-12-18 05:50:54 +00:00
StripSymbols	Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID	2013-01-01 13:57:25 +00:00
TailCallElim	Converted test dont-tce-tail-marked-call.ll to use FileCheck.	2013-01-11 04:16:35 +00:00
TailDup