[ValueTracking] Support signed intrinsic clamp

This is the same special logic we apply for SPF signed clamps
when computing the number of sign bits, just for intrinsics.

This just uses the same logic as the select case, but there's
multiple directions this could be improved in: We could also use
the num sign bits from the clamped value, we could do this during
constant range calculation, and there's probably unsigned analogues
for the constant range case at least.
This commit is contained in:
Nikita Popov 2022-02-23 12:40:31 +01:00
parent d6e008089c
commit 6777ec9e4d
2 changed files with 27 additions and 5 deletions

View File

@ -2902,6 +2902,24 @@ static bool isSignedMinMaxClamp(const Value *Select, const Value *&In,
return CLow->sle(*CHigh);
}
static bool isSignedMinMaxIntrinsicClamp(const IntrinsicInst *II,
const APInt *&CLow,
const APInt *&CHigh) {
assert((II->getIntrinsicID() == Intrinsic::smin ||
II->getIntrinsicID() == Intrinsic::smax) && "Must be smin/smax");
Intrinsic::ID InverseID = getInverseMinMaxIntrinsic(II->getIntrinsicID());
auto *InnerII = dyn_cast<IntrinsicInst>(II->getArgOperand(0));
if (!InnerII || InnerII->getIntrinsicID() != InverseID ||
!match(II->getArgOperand(1), m_APInt(CLow)) ||
!match(InnerII->getArgOperand(1), m_APInt(CHigh)))
return false;
if (II->getIntrinsicID() == Intrinsic::smin)
std::swap(CLow, CHigh);
return CLow->sle(*CHigh);
}
/// For vector constants, loop over the elements and find the constant with the
/// minimum number of sign bits. Return 0 if the value is not a vector constant
/// or if any element was not analyzed; otherwise, return the count for the
@ -3242,6 +3260,12 @@ static unsigned ComputeNumSignBitsImpl(const Value *V,
// Absolute value reduces number of sign bits by at most 1.
return Tmp - 1;
case Intrinsic::smin:
case Intrinsic::smax: {
const APInt *CLow, *CHigh;
if (isSignedMinMaxIntrinsicClamp(II, CLow, CHigh))
return std::min(CLow->getNumSignBits(), CHigh->getNumSignBits());
}
}
}
}

View File

@ -116,7 +116,7 @@ define i16 @min_max_clamp_intrinsic_2(i16 %x) {
; CHECK-LABEL: @min_max_clamp_intrinsic_2(
; CHECK-NEXT: [[A:%.*]] = call i16 @llvm.smin.i16(i16 [[X:%.*]], i16 2047)
; CHECK-NEXT: [[B:%.*]] = call i16 @llvm.smax.i16(i16 [[A]], i16 -2048)
; CHECK-NEXT: [[C:%.*]] = add i16 [[B]], 1
; CHECK-NEXT: [[C:%.*]] = add nsw i16 [[B]], 1
; CHECK-NEXT: ret i16 [[C]]
;
%a = call i16 @llvm.smin.i16(i16 %x, i16 2047)
@ -144,10 +144,8 @@ define i32 @min_max_clamp_intrinsic_4(i16 %x) {
; CHECK-LABEL: @min_max_clamp_intrinsic_4(
; CHECK-NEXT: [[A:%.*]] = call i16 @llvm.smin.i16(i16 [[X:%.*]], i16 2047)
; CHECK-NEXT: [[B:%.*]] = call i16 @llvm.smax.i16(i16 [[A]], i16 -2048)
; CHECK-NEXT: [[C:%.*]] = add i16 [[B]], 1
; CHECK-NEXT: [[D:%.*]] = sext i16 [[C]] to i32
; CHECK-NEXT: [[E:%.*]] = add nsw i32 [[D]], -1
; CHECK-NEXT: ret i32 [[E]]
; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[B]] to i32
; CHECK-NEXT: ret i32 [[TMP1]]
;
%a = call i16 @llvm.smin.i16(i16 %x, i16 2047)
%b = call i16 @llvm.smax.i16(i16 %a, i16 -2048)