This is an archive of the discontinued LLVM Phabricator instance.

Differential D65896

[LoopVectorize][X86] Clamp interleave factor if we have a known constant trip count that is less than VF*interleave
ClosedPublic

Authored by craig.topper on Aug 7 2019, 12:24 PM.

Details

Reviewers
hsaito
fhahn
karthiksenthil
Commits
rG005b22855e66: [LoopVectorize][X86] Clamp interleave factor if we have a known constant trip…
rL368215: [LoopVectorize][X86] Clamp interleave factor if we have a known constant trip…
Summary

If we know the trip count, we should make sure the interleave factor won't cause the vectorized loop to exceed it.

Improves one of the cases from PR42674

Diff Detail

Repository
rL LLVM

Event Timeline

craig.topper created this revision.Aug 7 2019, 12:24 PM
Herald added a project: Restricted Project. · View Herald TranscriptAug 7 2019, 12:24 PM
Herald added a subscriber: hiraditya. · View Herald Transcript
fhahn added inline comments.Aug 7 2019, 12:47 PM
llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
5139 ↗(On Diff #213966)

It might be slightly less code if we set MaxInterleaveCount = TC / VF; before line 5137, if TC > 0. TC / VF should always be >= 1. It might be worth adding an assertion to make sure we don't miss any cases.

craig.topper updated this revision to Diff 213977.Aug 7 2019, 1:11 PM

Factor TC/VF into MaxInterLeaveCount. Add assert that TC >= VF

fhahn accepted this revision.Aug 7 2019, 2:15 PM

LGTM, thanks.

This revision is now accepted and ready to land.Aug 7 2019, 2:15 PM
Closed by commit rL368215: [LoopVectorize][X86] Clamp interleave factor if we have a known constant trip… (authored by ctopper). · Explain WhyAug 7 2019, 2:44 PM
This revision was automatically updated to reflect the committed changes.

Revision Contents

PathSize
llvm/
trunk/
lib/
Transforms/
Vectorize/
10 lines
test/
Transforms/
LoopVectorize/
X86/
50 lines

Diff 214009

llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 5,115 Lines • ▼ Show 20 Linesunsigned LoopVectorizationCostModel::selectInterleaveCount(unsigned VF,
if (VF == 1) { if (VF == 1) {
if (ForceTargetMaxScalarInterleaveFactor.getNumOccurrences() > 0) if (ForceTargetMaxScalarInterleaveFactor.getNumOccurrences() > 0)
MaxInterleaveCount = ForceTargetMaxScalarInterleaveFactor; MaxInterleaveCount = ForceTargetMaxScalarInterleaveFactor;
} else { } else {
if (ForceTargetMaxVectorInterleaveFactor.getNumOccurrences() > 0) if (ForceTargetMaxVectorInterleaveFactor.getNumOccurrences() > 0)
MaxInterleaveCount = ForceTargetMaxVectorInterleaveFactor; MaxInterleaveCount = ForceTargetMaxVectorInterleaveFactor;
} }
// If the trip count is constant, limit the interleave count to be less than
// the trip count divided by VF.
if (TC > 0) {
assert(TC >= VF && "VF exceeds trip count?");
if ((TC / VF) < MaxInterleaveCount)
MaxInterleaveCount = (TC / VF);
}
// If we did not calculate the cost for VF (because the user selected the VF) // If we did not calculate the cost for VF (because the user selected the VF)
// then we calculate the cost of VF here. // then we calculate the cost of VF here.
if (LoopCost == 0) if (LoopCost == 0)
LoopCost = expectedCost(VF).first; LoopCost = expectedCost(VF).first;
assert(LoopCost && "Non-zero loop cost expected"); assert(LoopCost && "Non-zero loop cost expected");
// Clamp the calculated IC to be between the 1 and the max interleave count // Clamp the calculated IC to be between the 1 and the max interleave count
// that the target allows. // that the target and trip count allows.
if (IC > MaxInterleaveCount) if (IC > MaxInterleaveCount)
IC = MaxInterleaveCount; IC = MaxInterleaveCount;
else if (IC < 1) else if (IC < 1)
IC = 1; IC = 1;
// Interleave if we vectorized this loop and there is a reduction that could // Interleave if we vectorized this loop and there is a reduction that could
// benefit from interleaving. // benefit from interleaving.
if (VF > 1 && !Legal->getReductionVars()->empty()) { if (VF > 1 && !Legal->getReductionVars()->empty()) {
▲ Show 20 Lines • Show All 2,638 Lines • Show Last 20 Lines

llvm/trunk/test/Transforms/LoopVectorize/X86/pr42674.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt %s -loop-vectorize -instcombine -simplifycfg -mtriple=x86_64-unknown-linux-gnu -mcpu=skylake-avx512 -S | FileCheck %s ; RUN: opt %s -loop-vectorize -instcombine -simplifycfg -mtriple=x86_64-unknown-linux-gnu -mcpu=skylake-avx512 -S | FileCheck %s
@bytes = global [128 x i8] zeroinitializer, align 16 @bytes = global [128 x i8] zeroinitializer, align 16
; FIXME: We should end up with vector code for this loop, but don't because ; Make sure we end up with vector code for this loop. We used to try to create
; we try to create VF=64,UF=4 loop, but the scalar trip count is only 128 so ; a VF=64,UF=4 loop, but the scalar trip count is only 128 so
; the vector loop becomes dead code leaving only a scalar remainder. ; the vector loop was dead code leaving only a scalar remainder.
define zeroext i8 @sum() { define zeroext i8 @sum() {
; CHECK-LABEL: @sum( ; CHECK-LABEL: @sum(
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[FOR_BODY:%.*]] ; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: for.body: ; CHECK: vector.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[R_010:%.*]] = phi i8 [ 0, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] ; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <64 x i8> [ zeroinitializer, [[ENTRY]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [128 x i8], [128 x i8]* @bytes, i64 0, i64 [[INDVARS_IV]] ; CHECK-NEXT: [[VEC_PHI2:%.*]] = phi <64 x i8> [ zeroinitializer, [[ENTRY]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = load i8, i8* [[ARRAYIDX]], align 1 ; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [128 x i8], [128 x i8]* @bytes, i64 0, i64 [[INDEX]]
; CHECK-NEXT: [[ADD]] = add i8 [[TMP0]], [[R_010]] ; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8* [[TMP0]] to <64 x i8>*
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 ; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <64 x i8>, <64 x i8>* [[TMP1]], align 16
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 128 ; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP0]], i64 64
; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop !0 ; CHECK-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to <64 x i8>*
; CHECK: for.end: ; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <64 x i8>, <64 x i8>* [[TMP3]], align 16
; CHECK-NEXT: ret i8 [[ADD]] ; CHECK-NEXT: [[TMP4]] = add <64 x i8> [[WIDE_LOAD]], [[VEC_PHI]]
; CHECK-NEXT: [[TMP5]] = add <64 x i8> [[WIDE_LOAD3]], [[VEC_PHI2]]
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 128
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX]], 0
; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop !0
; CHECK: middle.block:
; CHECK-NEXT: [[BIN_RDX:%.*]] = add <64 x i8> [[TMP5]], [[TMP4]]
; CHECK-NEXT: [[RDX_SHUF:%.*]] = shufflevector <64 x i8> [[BIN_RDX]], <64 x i8> undef, <64 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
; CHECK-NEXT: [[BIN_RDX4:%.*]] = add <64 x i8> [[BIN_RDX]], [[RDX_SHUF]]
; CHECK-NEXT: [[RDX_SHUF5:%.*]] = shufflevector <64 x i8> [[BIN_RDX4]], <64 x i8> undef, <64 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
; CHECK-NEXT: [[BIN_RDX6:%.*]] = add <64 x i8> [[BIN_RDX4]], [[RDX_SHUF5]]
; CHECK-NEXT: [[RDX_SHUF7:%.*]] = shufflevector <64 x i8> [[BIN_RDX6]], <64 x i8> undef, <64 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
; CHECK-NEXT: [[BIN_RDX8:%.*]] = add <64 x i8> [[BIN_RDX6]], [[RDX_SHUF7]]
; CHECK-NEXT: [[RDX_SHUF9:%.*]] = shufflevector <64 x i8> [[BIN_RDX8]], <64 x i8> undef, <64 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
; CHECK-NEXT: [[BIN_RDX10:%.*]] = add <64 x i8> [[BIN_RDX8]], [[RDX_SHUF9]]
; CHECK-NEXT: [[RDX_SHUF11:%.*]] = shufflevector <64 x i8> [[BIN_RDX10]], <64 x i8> undef, <64 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
; CHECK-NEXT: [[BIN_RDX12:%.*]] = add <64 x i8> [[BIN_RDX10]], [[RDX_SHUF11]]
; CHECK-NEXT: [[RDX_SHUF13:%.*]] = shufflevector <64 x i8> [[BIN_RDX12]], <64 x i8> undef, <64 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
; CHECK-NEXT: [[BIN_RDX14:%.*]] = add <64 x i8> [[BIN_RDX12]], [[RDX_SHUF13]]
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <64 x i8> [[BIN_RDX14]], i32 0
; CHECK-NEXT: ret i8 [[TMP7]]
; ;
entry: entry:
br label %for.body br label %for.body
for.body: ; preds = %for.body, %entry for.body: ; preds = %for.body, %entry
%indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ] %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
%r.010 = phi i8 [ 0, %entry ], [ %add, %for.body ] %r.010 = phi i8 [ 0, %entry ], [ %add, %for.body ]
%arrayidx = getelementptr inbounds [128 x i8], [128 x i8]* @bytes, i64 0, i64 %indvars.iv %arrayidx = getelementptr inbounds [128 x i8], [128 x i8]* @bytes, i64 0, i64 %indvars.iv
Show All 10 Lines