Skip to content

[InstCombine] Fold minmax intrinsic using KnownBits information #76242

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Dec 22, 2023
Merged

[InstCombine] Fold minmax intrinsic using KnownBits information #76242

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bd763f6
[InstCombine] Add pre-commit tests. NFC.
dtcxzyw Dec 22, 2023
de719bd
[InstCombine] Fold minmax intrinsic using KnownBits information
dtcxzyw Dec 22, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions llvm/include/llvm/Analysis/ValueTracking.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -863,6 +863,11 @@ ConstantRange computeConstantRange(const Value *V, bool ForSigned,
const DominatorTree *DT = nullptr,
unsigned Depth = 0);

/// Combine constant ranges from computeConstantRange() and computeKnownBits().
ConstantRange
computeConstantRangeIncludingKnownBits(const WithCache<const Value *> &V,
bool ForSigned, const SimplifyQuery &SQ);

/// Return true if this function can prove that the instruction I will
/// always transfer execution to one of its successors (including the next
/// instruction that follows within a basic block). E.g. this is not
Expand Down
8 changes: 4 additions & 4 deletions llvm/lib/Analysis/ValueTracking.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6289,10 +6289,10 @@ static OverflowResult mapOverflowResult(ConstantRange::OverflowResult OR) {
}

/// Combine constant ranges from computeConstantRange() and computeKnownBits().
static ConstantRange
computeConstantRangeIncludingKnownBits(const WithCache<const Value *> &V,
bool ForSigned,
const SimplifyQuery &SQ) {
ConstantRange
llvm::computeConstantRangeIncludingKnownBits(const WithCache<const Value *> &V,
bool ForSigned,
const SimplifyQuery &SQ) {
ConstantRange CR1 =
ConstantRange::fromKnownBits(V.getKnownBits(SQ), ForSigned);
ConstantRange CR2 = computeConstantRange(V, ForSigned, SQ.IIQ.UseInstrInfo);
Expand Down
17 changes: 17 additions & 0 deletions llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1796,6 +1796,23 @@ Instruction *InstCombinerImpl::visitCallInst(CallInst &CI) {
if (Instruction *NewMinMax = factorizeMinMaxTree(II))
return NewMinMax;

// Try to fold minmax with constant RHS based on range information
const APInt *RHSC;
if (match(I1, m_APIntAllowUndef(RHSC))) {
ICmpInst::Predicate Pred =
ICmpInst::getNonStrictPredicate(MinMaxIntrinsic::getPredicate(IID));
bool IsSigned = MinMaxIntrinsic::isSigned(IID);
ConstantRange LHS_CR = computeConstantRangeIncludingKnownBits(
I0, IsSigned, SQ.getWithInstruction(II));
if (!LHS_CR.isFullSet()) {
if (LHS_CR.icmp(Pred, *RHSC))
return replaceInstUsesWith(*II, I0);
if (LHS_CR.icmp(ICmpInst::getSwappedPredicate(Pred), *RHSC))
return replaceInstUsesWith(*II,
ConstantInt::get(II->getType(), *RHSC));
}
}

break;
}
case Intrinsic::bitreverse: {
Expand Down
92 changes: 92 additions & 0 deletions llvm/test/Transforms/InstCombine/minmax-intrinsics.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2489,3 +2489,95 @@ define i1 @PR57986() {
%umin = call i1 @llvm.umin.i1(i1 ptrtoint (ptr @g to i1), i1 true)
ret i1 %umin
}

define i8 @fold_umax_with_knownbits_info(i8 %a, i8 %b) {
; CHECK-LABEL: @fold_umax_with_knownbits_info(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[A1:%.*]] = or i8 [[A:%.*]], 1
; CHECK-NEXT: [[A2:%.*]] = shl i8 [[B:%.*]], 1
; CHECK-NEXT: [[SUB:%.*]] = sub i8 [[A1]], [[A2]]
; CHECK-NEXT: ret i8 [[SUB]]
;
entry:
%a1 = or i8 %a, 1
%a2 = shl i8 %b, 1
%sub = sub i8 %a1, %a2
%val = call i8 @llvm.umax.i8(i8 %sub, i8 1)
ret i8 %val
}

define <3 x i8> @fold_umax_with_knownbits_info_undef_in_splat(<3 x i8> %a, <3 x i8> %b) {
; CHECK-LABEL: @fold_umax_with_knownbits_info_undef_in_splat(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[A1:%.*]] = or <3 x i8> [[A:%.*]], <i8 1, i8 1, i8 1>
; CHECK-NEXT: [[A2:%.*]] = shl <3 x i8> [[B:%.*]], <i8 1, i8 1, i8 1>
; CHECK-NEXT: [[SUB:%.*]] = sub <3 x i8> [[A1]], [[A2]]
; CHECK-NEXT: ret <3 x i8> [[SUB]]
;
entry:
%a1 = or <3 x i8> %a, <i8 1, i8 1, i8 1>
%a2 = shl <3 x i8> %b, <i8 1, i8 1, i8 1>
%sub = sub <3 x i8> %a1, %a2
%val = call <3 x i8> @llvm.umax.v3i8(<3 x i8> %sub, <3 x i8> <i8 1, i8 undef, i8 1>)
ret <3 x i8> %val
}

define i8 @fold_umin_with_knownbits_info(i8 %a, i8 %b) {
; CHECK-LABEL: @fold_umin_with_knownbits_info(
; CHECK-NEXT: entry:
; CHECK-NEXT: ret i8 3
;
entry:
%a1 = or i8 %a, 3
%a2 = shl i8 %b, 2
%sub = sub i8 %a1, %a2
%val = call i8 @llvm.umin.i8(i8 %sub, i8 3)
ret i8 %val
}

define <3 x i8> @fold_umin_with_knownbits_info_undef_in_splat(<3 x i8> %a, <3 x i8> %b) {
; CHECK-LABEL: @fold_umin_with_knownbits_info_undef_in_splat(
; CHECK-NEXT: entry:
; CHECK-NEXT: ret <3 x i8> <i8 3, i8 3, i8 3>
;
entry:
%a1 = or <3 x i8> %a, <i8 3, i8 3, i8 3>
%a2 = shl <3 x i8> %b, <i8 2, i8 2, i8 2>
%sub = sub <3 x i8> %a1, %a2
%val = call <3 x i8> @llvm.umin.v3i8(<3 x i8> %sub, <3 x i8> <i8 3, i8 undef, i8 3>)
ret <3 x i8> %val
}

define i8 @fold_umax_with_knownbits_info_fail(i8 %a, i8 %b) {
; CHECK-LABEL: @fold_umax_with_knownbits_info_fail(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[A1:%.*]] = or i8 [[A:%.*]], 2
; CHECK-NEXT: [[A2:%.*]] = shl i8 [[B:%.*]], 1
; CHECK-NEXT: [[SUB:%.*]] = sub i8 [[A1]], [[A2]]
; CHECK-NEXT: [[VAL:%.*]] = call i8 @llvm.umax.i8(i8 [[SUB]], i8 1)
; CHECK-NEXT: ret i8 [[VAL]]
;
entry:
%a1 = or i8 %a, 2
%a2 = shl i8 %b, 1
%sub = sub i8 %a1, %a2
%val = call i8 @llvm.umax.i8(i8 %sub, i8 1)
ret i8 %val
}

define i8 @fold_umin_with_knownbits_info_fail(i8 %a, i8 %b) {
; CHECK-LABEL: @fold_umin_with_knownbits_info_fail(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[A1:%.*]] = or i8 [[A:%.*]], 1
; CHECK-NEXT: [[A2:%.*]] = shl i8 [[B:%.*]], 2
; CHECK-NEXT: [[SUB:%.*]] = sub i8 [[A1]], [[A2]]
; CHECK-NEXT: [[VAL:%.*]] = call i8 @llvm.umin.i8(i8 [[SUB]], i8 3)
; CHECK-NEXT: ret i8 [[VAL]]
;
entry:
%a1 = or i8 %a, 1
%a2 = shl i8 %b, 2
%sub = sub i8 %a1, %a2
%val = call i8 @llvm.umin.i8(i8 %sub, i8 3)
ret i8 %val
}