Tim Northover 44697f3fc1 X86 CodeGenPrep: sink shufflevectors before shifts ...

On x86, shifting a vector by a scalar is significantly cheaper than shifting a
vector by another fully general vector. Unfortunately, because SelectionDAG
operates on just one basic block at a time, the shufflevector instruction that
reveals whether the right-hand side of a shift *is* really a scalar is often
not visible to CodeGen when it's needed.

This adds another handler to CodeGenPrepare, to sink any useful shufflevector
instructions down to the basic block where they're used, predicated on a target
hook (since on other architectures, doing so will often just introduce extra
real work).

rdar://problem/16063505

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@201655 91177308-0d34-0410-b5e6-96231b3b80d8

2014-02-19 10:02:43 +00:00

..

ADCE.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

CMakeLists.txt

Revert "Revert "Add Constant Hoisting Pass" (r200034)"

2014-01-25 02:02:55 +00:00

CodeGenPrepare.cpp

X86 CodeGenPrep: sink shufflevectors before shifts

2014-02-19 10:02:43 +00:00

ConstantHoisting.cpp

[Constant Hoisting] Fix insertion point for constant materialization.

2014-02-08 00:20:49 +00:00

ConstantProp.cpp

…

CorrelatedValuePropagation.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

DCE.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

DeadStoreElimination.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

EarlyCSE.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

FlattenCFGPass.cpp

…

GlobalMerge.cpp

GlobalMerge: move "-global-merge" option to the pass itself.

2014-02-18 11:17:29 +00:00

GVN.cpp

Make succ_iterator a real random access iterator and clean up a couple of users.

2014-02-10 14:17:42 +00:00

IndVarSimplify.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

JumpThreading.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LICM.cpp

[LPM] Switch LICM to actively use LCSSA in addition to preserving it.

2014-02-11 12:52:27 +00:00

LLVMBuild.txt

…

LoopDeletion.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LoopIdiomRecognize.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LoopInstSimplify.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LoopRerollPass.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LoopRotation.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LoopStrengthReduce.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LoopUnrollPass.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LoopUnswitch.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

LowerAtomic.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

Makefile

…

MemCpyOptimizer.cpp

A memcpy out of an fresh alloca is a no-op, delete it. Patch by Patrick Walton!

2014-02-06 06:29:19 +00:00

PartiallyInlineLibCalls.cpp

…

Reassociate.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

Reg2Mem.cpp

…

SampleProfile.cpp

[PM] Split DominatorTree into a concrete analysis result object which

2014-01-13 13:07:17 +00:00

Scalar.cpp

Revert "Revert "Add Constant Hoisting Pass" (r200034)"

2014-01-25 02:02:55 +00:00

Scalarizer.cpp

Fix Scalarizer insertion point when replacing PHIs with insertelements

2013-12-23 14:51:56 +00:00

ScalarReplAggregates.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

SCCP.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

SimplifyCFGPass.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00

Sink.cpp

[PM] Split DominatorTree into a concrete analysis result object which

2014-01-13 13:07:17 +00:00

SROA.cpp

Trivial cleanup: reuse existing variable.

2014-02-14 19:02:01 +00:00

StructurizeCFG.cpp

cleanup: scc_iterator consumers should use isAtEnd

2014-02-04 19:19:07 +00:00

TailRecursionElimination.cpp

Disable most IR-level transform passes on functions marked 'optnone'.

2014-02-06 00:07:05 +00:00