Simon Pilgrim 444b60212b [CostModel][AArch64] Add some initial costs for SK_Select and SK_PermuteSingleSrc ...

AArch64 was only setting costs for SK_Transpose, which meant that many of the simpler shuffles (e.g. SK_Select and SK_PermuteSingleSrc for larger vector elements) was being severely overestimated by the default shuffle expansion.

This patch adds costs to help improve SLP performance and avoid a regression in reductions introduced by D48174.

I'm not very knowledgeable about AArch64 shuffle lowering so I've kept the extra costs to a minimum - someone who knows this code can add extra costs which should improve vectorization a lot more.

Differential Revision: https://reviews.llvm.org/D48172

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@335329 91177308-0d34-0410-b5e6-96231b3b80d8

2018-06-22 09:45:31 +00:00

..

ADCE

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

AddDiscriminators

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

AggressiveInstCombine

[AggressiveInstCombine] avoid crashing on unsimplified code (PR37446)

2018-05-14 13:43:32 +00:00

AlignmentFromAssumptions

[AlignmentFromAssumptions] Set source and dest alignments of memory intrinsiscs separately

2018-02-22 18:55:59 +00:00

ArgumentPromotion

[ArgumentPromotion] don't break musttail invariant PR36543

2018-03-02 00:59:27 +00:00

AtomicExpand

Add address space mangling to lifetime intrinsics

2017-04-10 20:18:21 +00:00

BDCE

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

BlockExtractor

BlockExtractor: Don’t delete functions directly

2018-03-12 22:28:18 +00:00

BranchFolding

Move branch folding test to a better location.

2015-12-03 19:41:25 +00:00

CalledValuePropagation

Add CalledValuePropagation pass

2017-10-25 13:40:08 +00:00

CallSiteSplitting

[CallSiteSplit] Make sure we remove nonnull if the parameter turns out to be a constant.

2018-04-23 20:09:08 +00:00

CodeExtractor

[CodeExtractor] Allow extracting blocks with exception handling

2018-05-11 22:49:49 +00:00

CodeGenPrepare

Generalize MergeBlockIntoPredecessor. Replace uses of MergeBasicBlockIntoOnlyPred.

2018-06-20 22:01:04 +00:00

ConstantHoisting

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

ConstantMerge

Canonicalize the representation of empty an expression in DIGlobalVariableExpression

2017-08-30 18:06:51 +00:00

ConstProp

Make uitofp and sitofp defined on overflow.

2018-06-14 22:58:48 +00:00

Coroutines

[Coroutines] PR34897: Fix incorrect elisions

2018-05-11 03:12:28 +00:00

CorrelatedValuePropagation

[ValueTracking] Fix endless recursion in isKnownNonZero()

2018-05-30 15:56:46 +00:00

CrossDSOCFI

[cfi] Build __cfi_check as Thumb when applicable.

2017-08-29 22:29:15 +00:00

DCE

[DCE] Salvage debug info from dead insts

2018-02-15 22:26:18 +00:00

DeadArgElim

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

DeadStoreElimination

[Debugify] Move debug value intrinsics closer to their operand defs

2018-06-06 19:05:42 +00:00

DivRemPairs

[DivRemPairs] split tests per target to account for bots that don't build for all targets

2017-09-09 14:10:59 +00:00

EarlyCSE

[EarlyCSE] Propagate conditions of AND and OR instructions

2018-06-14 13:02:13 +00:00

EliminateAvailableExternally

[PM] Port EliminateAvailableExternally pass to the new pass manager.

2016-05-05 02:37:32 +00:00

EntryExitInstrumenter

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

ExpandMemCmp/X86

[x86, MemCmpExpansion] allow 2 pairs of loads per block (PR33325)

2018-01-06 16:16:04 +00:00

Float2Int

[PM] Port float2int to the new pass manager

2016-06-24 23:32:02 +00:00

ForcedFunctionAttrs

[attrs] Split off the forced attributes utility into its own pass that

2015-12-27 08:13:45 +00:00

FunctionAttrs

[CaptureTracking] Handle capturing of launder.invariant.group

2018-05-05 10:23:27 +00:00

FunctionImport

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

GCOVProfiling

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

GlobalDCE

[PM] Teach the PGO instrumentation pasess to run GlobalDCE before

2017-05-25 07:15:09 +00:00

GlobalMerge

[GlobalMerge] Set the alignment on merged global structs

2018-06-06 14:48:32 +00:00

GlobalOpt

Revert r335324 due to a builtbot failure

2018-06-22 08:57:01 +00:00

GlobalSplit

Fix one-after-the-end type metadata handling in globalsplit.

2017-03-07 22:18:48 +00:00

GuardWidening

[LoopGuardWidening] Make PostDomTree optional

2018-04-27 23:15:56 +00:00

GVN

[GVN] Avoid casting a vector of size less than 8 bits to i8

2018-06-21 21:43:20 +00:00

GVNHoist

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

GVNSink

Add an @llvm.sideeffect intrinsic

2017-11-08 21:59:51 +00:00

IndirectBrExpand

Introduce the "retpoline" x86 mitigation technique for variant #2 of the speculative execution vulnerabilities disclosed today, specifically identified by CVE-2017-5715, "Branch Target Injection", and is one of the two halves to Spectre..

2018-01-22 22:05:25 +00:00

IndVarSimplify

[SimplifyIndVars] Eliminate redundant truncs

2018-06-19 04:48:34 +00:00

InferAddressSpaces

[AMDGPU] Switch to the new addr space mapping by default

2018-02-02 16:07:16 +00:00

InferFunctionAttrs

[SimplifyLibcalls] Replace locked IO with unlocked IO

2018-05-16 11:39:52 +00:00

Inline

Fix aliasing of launder.invariant.group

2018-05-23 09:16:44 +00:00

InstCombine

[InstCombine] fix shuffle-of-binops bug

2018-06-21 23:56:59 +00:00

InstMerge

[MergedLoadStoreMotion] Fix a debug invariant bug in mergeStores

2018-05-09 06:52:12 +00:00

InstNamer

Add basic test case for -instnamer

2017-05-08 23:18:46 +00:00

InstSimplify

[InstSimplify] Fix missed optimization in simplifyUnsignedRangeCheck()

2018-06-20 14:22:49 +00:00

InterleavedAccess

[X86][LLVM]Expanding Supports lowerInterleaved{store|load}() in X86InterleavedAccess (VF64 stride 3-4)

2017-10-02 07:35:25 +00:00

Internalize

[Internalize] Test that __stack_chk_{guard, fail} are not internalized.

2016-06-05 19:08:54 +00:00

IPConstantProp

Revert r335206 "Recommit r333268: [IPSCCP] Use PredicateInfo to propagate facts from cmp instructions."

2018-06-21 19:18:36 +00:00

IRCE

[IRCE] Fix miscompile with range checks against negative values

2018-05-19 13:06:37 +00:00

JumpThreading

Fix the issue that ComputeValueKnownInPredecessors only handles the case when

2018-05-01 14:47:24 +00:00

LCSSA

[Debugify] Set a DI version module flag for llc compatibility

2018-05-24 23:00:23 +00:00

LICM

[MustExecute] Fix a debug invariant issue in isGuaranteedToExecute()

2018-05-25 13:02:59 +00:00

LoadStoreVectorizer

[AMDGPU] Re-enabled 128bit wide-vector generation for local addr space by default.

2018-05-28 18:15:11 +00:00

LoopDataPrefetch

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

LoopDeletion

[LoopDeletion] Handle users in unreachable block

2018-01-12 07:24:43 +00:00

LoopDistribute

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

LoopIdiom

[ValueTracking] Match select abs pattern when there's an sext involved

2018-06-04 16:53:57 +00:00

LoopInstSimplify

[LoopInstSimplify] Re-implement the core logic of loop-instsimplify to

2018-05-29 20:15:38 +00:00

LoopInterchange

[DA] Enable -da-delinearize by default

2018-06-21 11:53:16 +00:00

LoopLoadElim

[PM] Port LoopLoadElimination to the new pass manager and wire it into

2017-01-27 01:32:26 +00:00

LoopPredication

[LoopPredication] Add profitability check based on BPI

2018-03-22 16:03:59 +00:00

LoopReroll

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

LoopRotate

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

LoopSimplify

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

LoopSimplifyCFG

Generalize MergeBlockIntoPredecessor. Replace uses of MergeBasicBlockIntoOnlyPred.

2018-06-20 22:01:04 +00:00

LoopStrengthReduce

Generalize MergeBlockIntoPredecessor. Replace uses of MergeBasicBlockIntoOnlyPred.

2018-06-20 22:01:04 +00:00

LoopUnroll

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

LoopUnswitch

[LoopUnswitch] Fix SCEV invalidation in unswitching

2018-05-23 10:09:53 +00:00

LoopVectorize

[LoopVectorize] regenerate full checks; NFC

2018-06-21 16:54:32 +00:00

LoopVersioning

[LoopVersioning] Don't modify the list that we iterate over in addPHINodes

2018-05-22 08:33:02 +00:00

LoopVersioningLICM

[LAA] Correctly return a half-open range in expandBounds

2017-04-05 09:24:26 +00:00

LowerAtomic

LowerAtomic: Don't skip optnone functions; atomic still need lowering (PR34020)

2017-08-23 15:43:28 +00:00

LowerExpectIntrinsic

Fix builin_expect lowering bug

2017-06-07 18:32:24 +00:00

LowerGuardIntrinsic

[PM] Port LowerGuardIntrinsic to the new PM.

2016-07-28 22:08:41 +00:00

LowerInvoke

[PM] Port LowerInvoke to the new pass manager

2016-08-12 17:28:27 +00:00

LowerSwitch

LowerSwitch: Fix producing invalid IR on unreachable code

2017-04-21 23:54:12 +00:00

LowerTypeTests

In thin and full LTO + CFI, direct function calls may go through jump table

2018-06-04 18:18:12 +00:00

Mem2Reg

Re-apply "[DebugInfo] Check size of variable in ConvertDebugDeclareToDebugValue"

2018-06-15 13:48:55 +00:00

MemCpyOpt

[MemCpyOpt] Skip optimizing basic blocks not reachable from entry

2018-04-23 19:55:04 +00:00

MergeFunc

[MergeFunctions] Fix merging of small weak functions

2018-05-15 11:31:07 +00:00

MergeICmps

[MergeICmps] Don't crash when memcmp is not available

2018-05-19 12:51:59 +00:00

MetaRenamer

Remove alignment argument from memcpy/memmove/memset in favour of alignment attributes (Step 1)

2018-01-19 17:13:12 +00:00

NameAnonGlobals

[ThinLTO] Ensure anonymous globals renamed even at -O0

2016-09-17 20:40:16 +00:00

NaryReassociate

[NaryReassociate] Detect deleted instr with WeakVH

2018-05-24 06:09:02 +00:00

NewGVN

[NewGVN] Fix handling of assumes

2018-05-22 17:38:22 +00:00

ObjCARC

[ObjCARC] Prevent code motion into a catchswitch

2018-05-16 04:52:18 +00:00

PartiallyInlineLibCalls

[PartiallyInlineLibCalls][x86] add TTI hook to allow sqrt inlining to depend on arg rather than result

2017-11-27 21:15:43 +00:00

PGOProfile

[ICP] Do not attempt type matching for variable length arguments.

2018-04-25 17:19:21 +00:00

PhaseOrdering

[PhaseOrdering] remove stale comments; NFC

2018-05-09 23:10:46 +00:00

PlaceSafepoints

Remove alignment argument from memcpy/memmove/memset in favour of alignment attributes (Step 1)

2018-01-19 17:13:12 +00:00

PreISelIntrinsicLowering

[PM] Port PreISelIntrinsicLowering to the new PM

2016-06-24 20:13:42 +00:00

PruneEH

[PM][FunctionAttrs] add NoUnwind attribute inference to PostOrderFunctionAttrs pass

2018-03-23 21:46:16 +00:00

Reassociate

[Reassociate] Prevent infinite loops when processing PHIs.

2018-05-11 15:45:36 +00:00

Reg2Mem

Move the personality function from LandingPadInst to Function

2015-06-17 20:52:32 +00:00

RewriteStatepointsForGC

[RewriteStatepoints] Fix stale parse points

2018-03-05 22:27:30 +00:00

SafeStack

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

SampleProfile

[SampleFDO] Add a new compact binary format for sample profile.

2018-06-11 22:40:43 +00:00

Scalarizer

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

SCCP

Revert r335206 "Recommit r333268: [IPSCCP] Use PredicateInfo to propagate facts from cmp instructions."

2018-06-21 19:18:36 +00:00

SeparateConstOffsetFromGEP

[Split GEP] handle trunc() in separate-const-offset-from-gep pass.

2018-05-11 21:13:19 +00:00

SimpleLoopUnswitch

[LegacyPM] Fix PR37888 by teaching the legacy loop pass manager how to

2018-06-22 02:43:41 +00:00

SimplifyCFG

Revert "Simplify blockaddress usage before giving up in MergeBlockIntoPredecessor"

2018-06-18 23:20:08 +00:00

Sink

[Sink] Really really fix predicate in legality check

2018-01-11 21:28:57 +00:00

SLPVectorizer

[CostModel][AArch64] Add some initial costs for SK_Select and SK_PermuteSingleSrc

2018-06-22 09:45:31 +00:00

SpeculateAroundPHIs

Add a new pass to speculate around PHI nodes with constant (integer) operands when profitable.

2017-11-28 11:32:31 +00:00

SpeculativeExecution

Revert r301950: SpeculativeExecution: Stop using whitelist for costs

2017-05-10 12:30:07 +00:00

SROA

[SROA] Handle PHI with multiple duplicate predecessors

2018-05-17 07:21:41 +00:00

StraightLineStrengthReduce

AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel

2017-03-21 21:39:51 +00:00

StripDeadPrototypes

[PM] Port StripDeadPrototypes to the new pass manager

2015-10-30 23:28:12 +00:00

StripSymbols

Canonicalize the representation of empty an expression in DIGlobalVariableExpression

2017-08-30 18:06:51 +00:00

StructurizeCFG

StructurizeCFG: Adjust the loop depth for a subregion to order the nodes correctly

2018-05-23 18:34:48 +00:00

SyntheticCountsPropagation

Add a pass to generate synthetic function entry counts.

2018-01-09 19:39:35 +00:00

TailCallElim

[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label.

2018-05-09 02:40:45 +00:00

ThinLTOBitcodeWriter

[ThinLTOBitcodeWriter] Emit summaries for regular LTO modules

2018-06-01 15:20:47 +00:00

Util

[PredicateInfo] Order instructions in different BBs by DFSNumIn.

2018-06-20 17:42:01 +00:00

WholeProgramDevirt

Runtime flag to control branch funnel threshold

2018-04-06 21:32:36 +00:00