[InstCombine] Fold umax(X, C) + -C into usub.sat(X, C) (llvm#118195)

Alive2: https://alive2.llvm.org/ce/z/oSWe5S Closes llvm#118155
TIFitis · Dec 8, 2024 · 7c93bea · 7c93bea
1 parent a7c0833
commit 7c93bea
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 0 deletions.
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp b/llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp
@@ -994,6 +994,12 @@ Instruction *InstCombinerImpl::foldAddWithConstant(BinaryOperator &Add) {
     }
   }
 
+  // umax(X, C) + -C --> usub.sat(X, C)
+  if (match(Op0, m_OneUse(m_UMax(m_Value(X), m_SpecificInt(-*C)))))
+    return replaceInstUsesWith(
+        Add, Builder.CreateBinaryIntrinsic(
+                 Intrinsic::usub_sat, X, ConstantInt::get(Add.getType(), -*C)));
+
   // Fold (add (zext (add X, -1)), 1) -> (zext X) if X is non-zero.
   // TODO: There's a general form for any constant on the outer add.
   if (C->isOne()) {

diff --git a/llvm/test/Transforms/InstCombine/saturating-add-sub.ll b/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
@@ -2316,3 +2316,39 @@ define i32 @uadd_sat_via_add_swapped_cmp_select_nonstrict(i32 %x, i32 %y) {
   %r = select i1 %c, i32 %a, i32 -1
   ret i32 %r
 }
+
+define i8 @fold_add_umax_to_usub(i8 %a) {
+; CHECK-LABEL: @fold_add_umax_to_usub(
+; CHECK-NEXT:    [[SEL:%.*]] = call i8 @llvm.usub.sat.i8(i8 [[A:%.*]], i8 10)
+; CHECK-NEXT:    ret i8 [[SEL]]
+;
+  %umax = call i8 @llvm.umax.i8(i8 %a, i8 10)
+  %sel = add i8 %umax, -10
+  ret i8 %sel
+}
+
+define i8 @fold_add_umax_to_usub_incorrect_rhs(i8 %a) {
+; CHECK-LABEL: @fold_add_umax_to_usub_incorrect_rhs(
+; CHECK-NEXT:    [[UMAX:%.*]] = call i8 @llvm.umax.i8(i8 [[A:%.*]], i8 10)
+; CHECK-NEXT:    [[SEL:%.*]] = add i8 [[UMAX]], -11
+; CHECK-NEXT:    ret i8 [[SEL]]
+;
+  %umax = call i8 @llvm.umax.i8(i8 %a, i8 10)
+  %sel = add i8 %umax, -11
+  ret i8 %sel
+}
+
+define i8 @fold_add_umax_to_usub_multiuse(i8 %a) {
+; CHECK-LABEL: @fold_add_umax_to_usub_multiuse(
+; CHECK-NEXT:    [[UMAX:%.*]] = call i8 @llvm.umax.i8(i8 [[A:%.*]], i8 10)
+; CHECK-NEXT:    call void @usei8(i8 [[UMAX]])
+; CHECK-NEXT:    [[SEL:%.*]] = add i8 [[UMAX]], -10
+; CHECK-NEXT:    ret i8 [[SEL]]
+;
+  %umax = call i8 @llvm.umax.i8(i8 %a, i8 10)
+  call void @usei8(i8 %umax)
+  %sel = add i8 %umax, -10
+  ret i8 %sel
+}
+
+declare void @usei8(i8)