Skip to content

[InstCombine] Fold shift+cttz with power of 2 operands #127055

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 3 commits into from
Feb 18, 2025
Merged

[InstCombine] Fold shift+cttz with power of 2 operands #127055

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9f70feb
[InstCombine] Fold cttz with power of 2 operands
MDevereau Feb 13, 2025
c26fff8
Add fold for zero
MDevereau Feb 13, 2025
4c2004a
Remove unnecessary line
MDevereau Feb 17, 2025
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
16 changes: 16 additions & 0 deletions llvm/lib/Transforms/InstCombine/InstCombineShifts.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1613,6 +1613,22 @@ Instruction *InstCombinerImpl::visitLShr(BinaryOperator &I) {
if (Instruction *Overflow = foldLShrOverflowBit(I))
return Overflow;

// Transform ((pow2 << x) >> cttz(pow2 << y)) -> ((1 << x) >> y)
Value *Shl0_Op0, *Shl0_Op1, *Shl1_Op1;
BinaryOperator *Shl1;
if (match(Op0, m_Shl(m_Value(Shl0_Op0), m_Value(Shl0_Op1))) &&
match(Op1, m_Intrinsic<Intrinsic::cttz>(m_BinOp(Shl1))) &&
match(Shl1, m_Shl(m_Specific(Shl0_Op0), m_Value(Shl1_Op1))) &&
isKnownToBeAPowerOfTwo(Shl0_Op0, /*OrZero=*/true, 0, &I)) {
auto *Shl0 = cast<BinaryOperator>(Op0);
bool HasNUW = Shl0->hasNoUnsignedWrap() && Shl1->hasNoUnsignedWrap();
bool HasNSW = Shl0->hasNoSignedWrap() && Shl1->hasNoSignedWrap();
if (HasNUW || HasNSW) {
Value *NewShl = Builder.CreateShl(ConstantInt::get(Shl1->getType(), 1),
Shl0_Op1, "", HasNUW, HasNSW);
return BinaryOperator::CreateLShr(NewShl, Shl1_Op1);
}
}
return nullptr;
}

Expand Down
30 changes: 30 additions & 0 deletions llvm/test/Transforms/InstCombine/shift-cttz-ctlz.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -103,4 +103,34 @@ entry:
ret i32 %res
}

define i64 @fold_cttz_64() vscale_range(1,16) {
; CHECK-LABEL: define i64 @fold_cttz_64(
; CHECK-SAME: ) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry:
; CHECK-NEXT: ret i64 4
;
entry:
%vscale = tail call i64 @llvm.vscale.i64()
%shl0 = shl nuw nsw i64 %vscale, 4
%shl1 = shl nuw nsw i64 %vscale, 2
%cttz = tail call range(i64 2, 65) i64 @llvm.cttz.i64(i64 %shl1, i1 true)
%div1 = lshr i64 %shl0, %cttz
ret i64 %div1
}

define i32 @fold_cttz_32() vscale_range(1,16) {
; CHECK-LABEL: define i32 @fold_cttz_32(
; CHECK-SAME: ) #[[ATTR0]] {
; CHECK-NEXT: entry:
; CHECK-NEXT: ret i32 4
;
entry:
%vscale = tail call i32 @llvm.vscale.i32()
%shl0 = shl nuw nsw i32 %vscale, 4
%shl1 = shl nuw nsw i32 %vscale, 2
%cttz = tail call range(i32 2, 65) i32 @llvm.cttz.i32(i32 %shl1, i1 true)
%div1 = lshr i32 %shl0, %cttz
ret i32 %div1
}

declare void @use(i32)