[llvm] ede49fa - [InstCombine] Canonicalize more saturated-add variants (#100008)
Craig Topper via llvm-commits
llvm-commits at lists.llvm.org
Wed Aug 7 14:17:09 PDT 2024
Author: Rose
Date: 2024-08-07T14:16:50-07:00
New Revision: ede49fa5f3f32796618a71880038e2564c9d4db9
URL: https://github.com/llvm/llvm-project/commit/ede49fa5f3f32796618a71880038e2564c9d4db9
DIFF: https://github.com/llvm/llvm-project/commit/ede49fa5f3f32796618a71880038e2564c9d4db9.diff
LOG: [InstCombine] Canonicalize more saturated-add variants (#100008)
LLVM is not evaluating X u > C, a, b the same way it evaluates X <= C, b, a.
To fix this, let's move the folds to after the canonicalization of -1 to TrueVal.
Let's allow splat vectors with poison elements to be recognized too!
Finally, for completion, handle the one case that isn't caught by the above checks because it is canonicalized to eq:
X == -1 ? -1 : X + 1 -> uadd.sat(X, 1)
Alive2 Proof:
https://alive2.llvm.org/ce/z/WEcgYH
Added:
Modified:
llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
llvm/test/Transforms/InstCombine/saturating-add-sub.ll
Removed:
################################################################################
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
index 6025e73f07cf3..6bcff6f9d49ef 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
@@ -974,14 +974,7 @@ static Value *canonicalizeSaturatedAdd(ICmpInst *Cmp, Value *TVal, Value *FVal,
Value *Cmp1 = Cmp->getOperand(1);
ICmpInst::Predicate Pred = Cmp->getPredicate();
Value *X;
- const APInt *C, *CmpC;
- if (Pred == ICmpInst::ICMP_ULT &&
- match(TVal, m_Add(m_Value(X), m_APInt(C))) && X == Cmp0 &&
- match(FVal, m_AllOnes()) && match(Cmp1, m_APInt(CmpC)) && *CmpC == ~*C) {
- // (X u< ~C) ? (X + C) : -1 --> uadd.sat(X, C)
- return Builder.CreateBinaryIntrinsic(
- Intrinsic::uadd_sat, X, ConstantInt::get(X->getType(), *C));
- }
+ const APInt *C;
// Match unsigned saturated add of 2 variables with an unnecessary 'not'.
// There are 8 commuted variants.
@@ -993,6 +986,46 @@ static Value *canonicalizeSaturatedAdd(ICmpInst *Cmp, Value *TVal, Value *FVal,
if (!match(TVal, m_AllOnes()))
return nullptr;
+ // uge -1 is canonicalized to eq -1 and requires special handling
+ // (a == -1) ? -1 : a + 1 -> uadd.sat(a, 1)
+ if (Pred == ICmpInst::ICMP_EQ) {
+ if (match(FVal, m_Add(m_Specific(Cmp0), m_One())) &&
+ match(Cmp1, m_AllOnes())) {
+ return Builder.CreateBinaryIntrinsic(
+ Intrinsic::uadd_sat, Cmp0, ConstantInt::get(Cmp0->getType(), 1));
+ }
+ return nullptr;
+ }
+
+ if ((Pred == ICmpInst::ICMP_UGE || Pred == ICmpInst::ICMP_UGT) &&
+ match(FVal, m_Add(m_Specific(Cmp0), m_APIntAllowPoison(C))) &&
+ match(Cmp1, m_SpecificIntAllowPoison(~*C))) {
+ // (X u> ~C) ? -1 : (X + C) --> uadd.sat(X, C)
+ // (X u>= ~C)? -1 : (X + C) --> uadd.sat(X, C)
+ return Builder.CreateBinaryIntrinsic(Intrinsic::uadd_sat, Cmp0,
+ ConstantInt::get(Cmp0->getType(), *C));
+ }
+
+ // Negative one does not work here because X u> -1 ? -1, X + -1 is not a
+ // saturated add.
+ if (Pred == ICmpInst::ICMP_UGT &&
+ match(FVal, m_Add(m_Specific(Cmp0), m_APIntAllowPoison(C))) &&
+ match(Cmp1, m_SpecificIntAllowPoison(~*C - 1)) && !C->isAllOnes()) {
+ // (X u> ~C - 1) ? -1 : (X + C) --> uadd.sat(X, C)
+ return Builder.CreateBinaryIntrinsic(Intrinsic::uadd_sat, Cmp0,
+ ConstantInt::get(Cmp0->getType(), *C));
+ }
+
+ // Zero does not work here because X u>= 0 ? -1 : X -> is always -1, which is
+ // not a saturated add.
+ if (Pred == ICmpInst::ICMP_UGE &&
+ match(FVal, m_Add(m_Specific(Cmp0), m_APIntAllowPoison(C))) &&
+ match(Cmp1, m_SpecificIntAllowPoison(-*C)) && !C->isZero()) {
+ // (X u >= -C) ? -1 : (X + C) --> uadd.sat(X, C)
+ return Builder.CreateBinaryIntrinsic(Intrinsic::uadd_sat, Cmp0,
+ ConstantInt::get(Cmp0->getType(), *C));
+ }
+
// Canonicalize predicate to less-than or less-or-equal-than.
if (Pred == ICmpInst::ICMP_UGT || Pred == ICmpInst::ICMP_UGE) {
std::swap(Cmp0, Cmp1);
diff --git a/llvm/test/Transforms/InstCombine/saturating-add-sub.ll b/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
index 87ccce566edca..d23f8d48e0c71 100644
--- a/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
+++ b/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
@@ -1398,9 +1398,7 @@ define i32 @uadd_sat(i32 %x, i32 %y) {
define i32 @uadd_sat_flipped(i32 %x) {
; CHECK-LABEL: @uadd_sat_flipped(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i32 [[X:%.*]], -11
-; CHECK-NEXT: [[ADD:%.*]] = add i32 [[X]], 9
-; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP]], i32 -1, i32 [[ADD]]
+; CHECK-NEXT: [[COND:%.*]] = call i32 @llvm.uadd.sat.i32(i32 [[X:%.*]], i32 9)
; CHECK-NEXT: ret i32 [[COND]]
;
%cmp = icmp ugt i32 %x, -11
@@ -1411,9 +1409,7 @@ define i32 @uadd_sat_flipped(i32 %x) {
define i32 @uadd_sat_flipped2(i32 %x) {
; CHECK-LABEL: @uadd_sat_flipped2(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i32 [[X:%.*]], -10
-; CHECK-NEXT: [[ADD:%.*]] = add i32 [[X]], 9
-; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP]], i32 -1, i32 [[ADD]]
+; CHECK-NEXT: [[COND:%.*]] = call i32 @llvm.uadd.sat.i32(i32 [[X:%.*]], i32 9)
; CHECK-NEXT: ret i32 [[COND]]
;
%cmp = icmp ugt i32 %x, -10
@@ -1452,9 +1448,7 @@ define i32 @uadd_sat_flipped3_neg_no_nuw(i32 %x) {
define i32 @uadd_sat_negative_one(i32 %x) {
; CHECK-LABEL: @uadd_sat_negative_one(
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[X:%.*]], -1
-; CHECK-NEXT: [[ADD:%.*]] = add i32 [[X]], 1
-; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP]], i32 -1, i32 [[ADD]]
+; CHECK-NEXT: [[COND:%.*]] = call i32 @llvm.uadd.sat.i32(i32 [[X:%.*]], i32 1)
; CHECK-NEXT: ret i32 [[COND]]
;
%cmp = icmp eq i32 %x, -1
@@ -1476,9 +1470,7 @@ define <2 x i8> @uadd_sat_flipped4_vector(<2 x i8> %x) {
define <2 x i8> @uadd_sat_flipped4_poison_vector(<2 x i8> %x) {
; CHECK-LABEL: @uadd_sat_flipped4_poison_vector(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ult <2 x i8> [[X:%.*]], <i8 -10, i8 poison>
-; CHECK-NEXT: [[ADD:%.*]] = add <2 x i8> [[X]], <i8 9, i8 9>
-; CHECK-NEXT: [[COND:%.*]] = select <2 x i1> [[CMP]], <2 x i8> [[ADD]], <2 x i8> <i8 -1, i8 -1>
+; CHECK-NEXT: [[COND:%.*]] = call <2 x i8> @llvm.uadd.sat.v2i8(<2 x i8> [[X:%.*]], <2 x i8> <i8 9, i8 9>)
; CHECK-NEXT: ret <2 x i8> [[COND]]
;
%cmp = icmp ult <2 x i8> %x, <i8 -10, i8 poison>
@@ -1489,9 +1481,7 @@ define <2 x i8> @uadd_sat_flipped4_poison_vector(<2 x i8> %x) {
define <2 x i8> @uadd_sat_flipped4_poison_vector_compare(<2 x i8> %x) {
; CHECK-LABEL: @uadd_sat_flipped4_poison_vector_compare(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ult <2 x i8> [[X:%.*]], <i8 -10, i8 poison>
-; CHECK-NEXT: [[ADD:%.*]] = add <2 x i8> [[X]], <i8 9, i8 poison>
-; CHECK-NEXT: [[COND:%.*]] = select <2 x i1> [[CMP]], <2 x i8> [[ADD]], <2 x i8> <i8 -1, i8 -1>
+; CHECK-NEXT: [[COND:%.*]] = call <2 x i8> @llvm.uadd.sat.v2i8(<2 x i8> [[X:%.*]], <2 x i8> <i8 9, i8 9>)
; CHECK-NEXT: ret <2 x i8> [[COND]]
;
%cmp = icmp ult <2 x i8> %x, <i8 -10, i8 poison>
@@ -1986,9 +1976,7 @@ define i32 @uadd_sat_not_commute_select_uge_commute_add(i32 %x, i32 %y) {
define i32 @uadd_sat_constant(i32 %x) {
; CHECK-LABEL: @uadd_sat_constant(
-; CHECK-NEXT: [[A:%.*]] = add i32 [[X:%.*]], 42
-; CHECK-NEXT: [[C:%.*]] = icmp ugt i32 [[X]], -43
-; CHECK-NEXT: [[R:%.*]] = select i1 [[C]], i32 -1, i32 [[A]]
+; CHECK-NEXT: [[R:%.*]] = call i32 @llvm.uadd.sat.i32(i32 [[X:%.*]], i32 42)
; CHECK-NEXT: ret i32 [[R]]
;
%a = add i32 %x, 42
@@ -2054,9 +2042,7 @@ define i32 @uadd_sat_canon_y_nuw(i32 %x, i32 %y) {
define <4 x i32> @uadd_sat_constant_vec(<4 x i32> %x) {
; CHECK-LABEL: @uadd_sat_constant_vec(
-; CHECK-NEXT: [[A:%.*]] = add <4 x i32> [[X:%.*]], <i32 42, i32 42, i32 42, i32 42>
-; CHECK-NEXT: [[C:%.*]] = icmp ugt <4 x i32> [[X]], <i32 -43, i32 -43, i32 -43, i32 -43>
-; CHECK-NEXT: [[R:%.*]] = select <4 x i1> [[C]], <4 x i32> <i32 -1, i32 -1, i32 -1, i32 -1>, <4 x i32> [[A]]
+; CHECK-NEXT: [[R:%.*]] = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> [[X:%.*]], <4 x i32> <i32 42, i32 42, i32 42, i32 42>)
; CHECK-NEXT: ret <4 x i32> [[R]]
;
%a = add <4 x i32> %x, <i32 42, i32 42, i32 42, i32 42>
@@ -2078,9 +2064,7 @@ define <4 x i32> @uadd_sat_constant_vec_commute(<4 x i32> %x) {
define <4 x i32> @uadd_sat_constant_vec_commute_undefs(<4 x i32> %x) {
; CHECK-LABEL: @uadd_sat_constant_vec_commute_undefs(
-; CHECK-NEXT: [[A:%.*]] = add <4 x i32> [[X:%.*]], <i32 42, i32 42, i32 42, i32 poison>
-; CHECK-NEXT: [[C:%.*]] = icmp ult <4 x i32> [[X]], <i32 -43, i32 -43, i32 poison, i32 -43>
-; CHECK-NEXT: [[R:%.*]] = select <4 x i1> [[C]], <4 x i32> [[A]], <4 x i32> <i32 -1, i32 poison, i32 -1, i32 -1>
+; CHECK-NEXT: [[R:%.*]] = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> [[X:%.*]], <4 x i32> <i32 42, i32 42, i32 42, i32 42>)
; CHECK-NEXT: ret <4 x i32> [[R]]
;
%a = add <4 x i32> %x, <i32 42, i32 42, i32 42, i32 poison>
More information about the llvm-commits
mailing list