[ValueTracking] Add support for non-splat vecs in computeConstantRange #72365
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
llvmbot
added
llvm:analysis
Member
|
@llvm/pr-subscribers-llvm-analysis Author: Yingwei Zheng (dtcxzyw) ChangesRelated patch: #68331 Full diff: https://github.com/llvm/llvm-project/pull/72365.diff 4 Files Affected:
diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp
index dc3c90ba9db1664..5f5d7e07cac1e46 100644
--- a/llvm/lib/Analysis/ValueTracking.cpp
+++ b/llvm/lib/Analysis/ValueTracking.cpp
@@ -8871,9 +8871,17 @@ ConstantRange llvm::computeConstantRange(const Value *V, bool ForSigned,
const APInt *C;
if (match(V, m_APInt(C)))
return ConstantRange(*C);
+ unsigned BitWidth = V->getType()->getScalarSizeInBits();
+
+ if (auto *VC = dyn_cast<ConstantDataVector>(V)) {
+ ConstantRange CR = ConstantRange::getEmpty(BitWidth);
+ for (unsigned ElemIdx = 0, NElem = VC->getNumElements(); ElemIdx < NElem;
+ ++ElemIdx)
+ CR = CR.unionWith(VC->getElementAsAPInt(ElemIdx));
+ return CR;
+ }
InstrInfoQuery IIQ(UseInstrInfo);
- unsigned BitWidth = V->getType()->getScalarSizeInBits();
ConstantRange CR = ConstantRange::getFull(BitWidth);
if (auto *BO = dyn_cast<BinaryOperator>(V)) {
APInt Lower = APInt(BitWidth, 0);
diff --git a/llvm/test/Transforms/InstCombine/add.ll b/llvm/test/Transforms/InstCombine/add.ll
index 1079f4fab1cd8e1..2ae0181974bf8a6 100644
--- a/llvm/test/Transforms/InstCombine/add.ll
+++ b/llvm/test/Transforms/InstCombine/add.ll
@@ -440,11 +440,9 @@ define <2 x i8> @test18vec_nsw(<2 x i8> %A) {
ret <2 x i8> %C
}
-; TODO: fix ValueTracking overflow check for non-splat vector, could be attached nsw
-; this shouldn't overflow.
define <2 x i8> @test18vec_nsw_false(<2 x i8> %A) {
; CHECK-LABEL: @test18vec_nsw_false(
-; CHECK-NEXT: [[C:%.*]] = sub <2 x i8> <i8 -125, i8 -126>, [[A:%.*]]
+; CHECK-NEXT: [[C:%.*]] = sub nsw <2 x i8> <i8 -125, i8 -126>, [[A:%.*]]
; CHECK-NEXT: ret <2 x i8> [[C]]
;
%B = xor <2 x i8> %A, <i8 -1, i8 -1>
diff --git a/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll b/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll
index 83a45e697e59e16..3dbd80c773d141d 100644
--- a/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll
+++ b/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll
@@ -207,7 +207,7 @@ define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov1(<2 x i8> %arg
; TODO: We can add nsw on sub, current Value Tracking use [max element,min element] constant range, to check overflow for vector?
define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov2(<2 x i8> %arg) {
; CHECK-LABEL: @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov2(
-; CHECK-NEXT: [[T1:%.*]] = sub <2 x i8> <i8 -126, i8 -128>, [[ARG:%.*]]
+; CHECK-NEXT: [[T1:%.*]] = sub nsw <2 x i8> <i8 -126, i8 -128>, [[ARG:%.*]]
; CHECK-NEXT: ret <2 x i8> [[T1]]
;
%t0 = add nsw <2 x i8> %arg, <i8 1, i8 2>
@@ -218,7 +218,7 @@ define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov2(<2 x i8> %arg
; TODO: We can add nsw on sub, curret Value Tracking can't decide this is not overflowed?
define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov3(<2 x i8> %arg) {
; CHECK-LABEL: @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov3(
-; CHECK-NEXT: [[T1:%.*]] = sub <2 x i8> <i8 -120, i8 -127>, [[ARG:%.*]]
+; CHECK-NEXT: [[T1:%.*]] = sub nsw <2 x i8> <i8 -120, i8 -127>, [[ARG:%.*]]
; CHECK-NEXT: ret <2 x i8> [[T1]]
;
%t0 = add nsw <2 x i8> %arg, <i8 0, i8 1>
diff --git a/llvm/test/Transforms/InstCombine/saturating-add-sub.ll b/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
index 612c5bf77462d4b..c1bb6941d456836 100644
--- a/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
+++ b/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
@@ -1052,11 +1052,9 @@ define <2 x i8> @test_vector_usub_add_nuw_no_ov(<2 x i8> %a) {
ret <2 x i8> %r
}
-; Can be optimized if the usub.sat RHS constant range handles non-splat vectors.
define <2 x i8> @test_vector_usub_add_nuw_no_ov_nonsplat1(<2 x i8> %a) {
; CHECK-LABEL: @test_vector_usub_add_nuw_no_ov_nonsplat1(
-; CHECK-NEXT: [[B:%.*]] = add nuw <2 x i8> [[A:%.*]], <i8 10, i8 10>
-; CHECK-NEXT: [[R:%.*]] = call <2 x i8> @llvm.usub.sat.v2i8(<2 x i8> [[B]], <2 x i8> <i8 10, i8 9>)
+; CHECK-NEXT: [[R:%.*]] = add <2 x i8> [[A:%.*]], <i8 0, i8 1>
; CHECK-NEXT: ret <2 x i8> [[R]]
;
%b = add nuw <2 x i8> %a, <i8 10, i8 10>
|
Member
|
@llvm/pr-subscribers-llvm-transforms Author: Yingwei Zheng (dtcxzyw) ChangesRelated patch: #68331 Full diff: https://github.com/llvm/llvm-project/pull/72365.diff 4 Files Affected:
diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp
index dc3c90ba9db1664..5f5d7e07cac1e46 100644
--- a/llvm/lib/Analysis/ValueTracking.cpp
+++ b/llvm/lib/Analysis/ValueTracking.cpp
@@ -8871,9 +8871,17 @@ ConstantRange llvm::computeConstantRange(const Value *V, bool ForSigned,
const APInt *C;
if (match(V, m_APInt(C)))
return ConstantRange(*C);
+ unsigned BitWidth = V->getType()->getScalarSizeInBits();
+
+ if (auto *VC = dyn_cast<ConstantDataVector>(V)) {
+ ConstantRange CR = ConstantRange::getEmpty(BitWidth);
+ for (unsigned ElemIdx = 0, NElem = VC->getNumElements(); ElemIdx < NElem;
+ ++ElemIdx)
+ CR = CR.unionWith(VC->getElementAsAPInt(ElemIdx));
+ return CR;
+ }
InstrInfoQuery IIQ(UseInstrInfo);
- unsigned BitWidth = V->getType()->getScalarSizeInBits();
ConstantRange CR = ConstantRange::getFull(BitWidth);
if (auto *BO = dyn_cast<BinaryOperator>(V)) {
APInt Lower = APInt(BitWidth, 0);
diff --git a/llvm/test/Transforms/InstCombine/add.ll b/llvm/test/Transforms/InstCombine/add.ll
index 1079f4fab1cd8e1..2ae0181974bf8a6 100644
--- a/llvm/test/Transforms/InstCombine/add.ll
+++ b/llvm/test/Transforms/InstCombine/add.ll
@@ -440,11 +440,9 @@ define <2 x i8> @test18vec_nsw(<2 x i8> %A) {
ret <2 x i8> %C
}
-; TODO: fix ValueTracking overflow check for non-splat vector, could be attached nsw
-; this shouldn't overflow.
define <2 x i8> @test18vec_nsw_false(<2 x i8> %A) {
; CHECK-LABEL: @test18vec_nsw_false(
-; CHECK-NEXT: [[C:%.*]] = sub <2 x i8> <i8 -125, i8 -126>, [[A:%.*]]
+; CHECK-NEXT: [[C:%.*]] = sub nsw <2 x i8> <i8 -125, i8 -126>, [[A:%.*]]
; CHECK-NEXT: ret <2 x i8> [[C]]
;
%B = xor <2 x i8> %A, <i8 -1, i8 -1>
diff --git a/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll b/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll
index 83a45e697e59e16..3dbd80c773d141d 100644
--- a/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll
+++ b/llvm/test/Transforms/InstCombine/addsub-constant-folding.ll
@@ -207,7 +207,7 @@ define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov1(<2 x i8> %arg
; TODO: We can add nsw on sub, current Value Tracking use [max element,min element] constant range, to check overflow for vector?
define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov2(<2 x i8> %arg) {
; CHECK-LABEL: @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov2(
-; CHECK-NEXT: [[T1:%.*]] = sub <2 x i8> <i8 -126, i8 -128>, [[ARG:%.*]]
+; CHECK-NEXT: [[T1:%.*]] = sub nsw <2 x i8> <i8 -126, i8 -128>, [[ARG:%.*]]
; CHECK-NEXT: ret <2 x i8> [[T1]]
;
%t0 = add nsw <2 x i8> %arg, <i8 1, i8 2>
@@ -218,7 +218,7 @@ define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov2(<2 x i8> %arg
; TODO: We can add nsw on sub, curret Value Tracking can't decide this is not overflowed?
define <2 x i8> @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov3(<2 x i8> %arg) {
; CHECK-LABEL: @non_splat_vec_add_nsw_const_const_sub_nsw_not_ov3(
-; CHECK-NEXT: [[T1:%.*]] = sub <2 x i8> <i8 -120, i8 -127>, [[ARG:%.*]]
+; CHECK-NEXT: [[T1:%.*]] = sub nsw <2 x i8> <i8 -120, i8 -127>, [[ARG:%.*]]
; CHECK-NEXT: ret <2 x i8> [[T1]]
;
%t0 = add nsw <2 x i8> %arg, <i8 0, i8 1>
diff --git a/llvm/test/Transforms/InstCombine/saturating-add-sub.ll b/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
index 612c5bf77462d4b..c1bb6941d456836 100644
--- a/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
+++ b/llvm/test/Transforms/InstCombine/saturating-add-sub.ll
@@ -1052,11 +1052,9 @@ define <2 x i8> @test_vector_usub_add_nuw_no_ov(<2 x i8> %a) {
ret <2 x i8> %r
}
-; Can be optimized if the usub.sat RHS constant range handles non-splat vectors.
define <2 x i8> @test_vector_usub_add_nuw_no_ov_nonsplat1(<2 x i8> %a) {
; CHECK-LABEL: @test_vector_usub_add_nuw_no_ov_nonsplat1(
-; CHECK-NEXT: [[B:%.*]] = add nuw <2 x i8> [[A:%.*]], <i8 10, i8 10>
-; CHECK-NEXT: [[R:%.*]] = call <2 x i8> @llvm.usub.sat.v2i8(<2 x i8> [[B]], <2 x i8> <i8 10, i8 9>)
+; CHECK-NEXT: [[R:%.*]] = add <2 x i8> [[A:%.*]], <i8 0, i8 1>
; CHECK-NEXT: ret <2 x i8> [[R]]
;
%b = add nuw <2 x i8> %a, <i8 10, i8 10>
|
zahiraam
pushed a commit
to zahiraam/llvm-project
that referenced
this pull request
Nov 20, 2023
llvm#72365) Related patch: llvm#68331 This missed optimization is discovered with the help of AliveToolkit/alive2#962.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Related patch: #68331
This missed optimization is discovered with the help of AliveToolkit/alive2#962.