[InstCombine] Fold fcmp ord/uno (fptrunc X), C to fcmp ord/uno X, C - #185848
Closed
wermos wants to merge 4 commits into
Closed
[InstCombine] Fold fcmp ord/uno (fptrunc X), C to fcmp ord/uno X, C#185848wermos wants to merge 4 commits into
fcmp ord/uno (fptrunc X), C to fcmp ord/uno X, C#185848wermos wants to merge 4 commits into
Conversation
llvmbot
added
llvm:instcombine
Member
|
@llvm/pr-subscribers-llvm-transforms Author: Tirthankar Mazumder (wermos) ChangesFixes #185698. Alive2 proof: https://alive2.llvm.org/ce/z/682cYT Full diff: https://github.com/llvm/llvm-project/pull/185848.diff 2 Files Affected:
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 10caf4ab78a5a..90e9fbb4ab583 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -8359,26 +8359,15 @@ static Instruction *foldFCmpReciprocalAndZero(FCmpInst &I, Instruction *LHSI,
// Transform 'fptrunc(x) cmp C' to 'x cmp ext(C)' if possible.
// Patterns include:
-// fptrunc(x) < C --> x < ext(C)
-// fptrunc(x) <= C --> x <= ext(C)
-// fptrunc(x) > C --> x > ext(C)
-// fptrunc(x) >= C --> x >= ext(C)
+// fptrunc(x) < C --> x < ext(C)
+// fptrunc(x) <= C --> x <= ext(C)
+// fptrunc(x) > C --> x > ext(C)
+// fptrunc(x) >= C --> x >= ext(C)
+// fptrunc(x) ord/uno C --> x ord/uno C
// where 'ext(C)' is the extension of 'C' to the type of 'x' with a small bias
// due to precision loss.
static Instruction *foldFCmpFpTrunc(FCmpInst &I, const Instruction &FPTrunc,
const Constant &C) {
- FCmpInst::Predicate Pred = I.getPredicate();
- bool RoundDown = false;
-
- if (Pred == FCmpInst::FCMP_OGE || Pred == FCmpInst::FCMP_UGE ||
- Pred == FCmpInst::FCMP_OLT || Pred == FCmpInst::FCMP_ULT)
- RoundDown = true;
- else if (Pred == FCmpInst::FCMP_OGT || Pred == FCmpInst::FCMP_UGT ||
- Pred == FCmpInst::FCMP_OLE || Pred == FCmpInst::FCMP_ULE)
- RoundDown = false;
- else
- return nullptr;
-
const APFloat *CValue;
if (!match(&C, m_APFloat(CValue)))
return nullptr;
@@ -8393,6 +8382,31 @@ static Instruction *foldFCmpFpTrunc(FCmpInst &I, const Instruction &FPTrunc,
return Dest;
};
+ Type *DestType = FPTrunc.getOperand(0)->getType();
+ const fltSemantics &DestFltSema =
+ DestType->getScalarType()->getFltSemantics();
+
+ APFloat ExtCValue = ConvertFltSema(*CValue, DestFltSema);
+
+ FCmpInst::Predicate Pred = I.getPredicate();
+
+ // Fold fcmp ord/uno fptrunc X, C -> fcmp ord/uno X, C
+ if (Pred == FCmpInst::FCMP_ORD || Pred == FCmpInst::FCMP_UNO) {
+ return new FCmpInst(Pred, FPTrunc.getOperand(0),
+ ConstantFP::get(DestType, ExtCValue), "", &I);
+ }
+
+ bool RoundDown = false;
+
+ if (Pred == FCmpInst::FCMP_OGE || Pred == FCmpInst::FCMP_UGE ||
+ Pred == FCmpInst::FCMP_OLT || Pred == FCmpInst::FCMP_ULT)
+ RoundDown = true;
+ else if (Pred == FCmpInst::FCMP_OGT || Pred == FCmpInst::FCMP_UGT ||
+ Pred == FCmpInst::FCMP_OLE || Pred == FCmpInst::FCMP_ULE)
+ RoundDown = false;
+ else
+ return nullptr;
+
auto NextValue = [](const APFloat &Value, bool RoundDown) {
APFloat NextValue = Value;
NextValue.next(RoundDown);
@@ -8400,12 +8414,6 @@ static Instruction *foldFCmpFpTrunc(FCmpInst &I, const Instruction &FPTrunc,
};
APFloat NextCValue = NextValue(*CValue, RoundDown);
-
- Type *DestType = FPTrunc.getOperand(0)->getType();
- const fltSemantics &DestFltSema =
- DestType->getScalarType()->getFltSemantics();
-
- APFloat ExtCValue = ConvertFltSema(*CValue, DestFltSema);
APFloat ExtNextCValue = ConvertFltSema(NextCValue, DestFltSema);
// When 'NextCValue' is infinity, use an imaged 'NextCValue' that equals
diff --git a/llvm/test/Transforms/InstCombine/fold-fcmp-trunc.ll b/llvm/test/Transforms/InstCombine/fold-fcmp-trunc.ll
index 371f9b6807fe4..7bc5d7fb7a1c6 100644
--- a/llvm/test/Transforms/InstCombine/fold-fcmp-trunc.ll
+++ b/llvm/test/Transforms/InstCombine/fold-fcmp-trunc.ll
@@ -104,7 +104,7 @@ define i1 @fcmp_trunc_with_reassoc(double %0) {
define i1 @fcmp_trunc_with_fast(double %0) {
; CHECK-LABEL: define i1 @fcmp_trunc_with_fast(
; CHECK-SAME: double [[TMP0:%.*]]) {
-; CHECK-NEXT: [[RESULT:%.*]] = fcmp fast oge double [[TMP0]], 0x4058FFFFF0000000
+; CHECK-NEXT: [[RESULT:%.*]] = fcmp fast oge double [[TMP0]], 0x4058FFFFF0000000
; CHECK-NEXT: ret i1 [[RESULT]]
;
%trunc = fptrunc double %0 to float
@@ -672,3 +672,50 @@ define i1 @fcmp_trunc_mn_ppc_fp128(ppc_fp128 %0) {
ret i1 %result
}
+define i1 @fptrunc_uno_fcmp(double %arg0) {
+; CHECK-LABEL: define i1 @fptrunc_uno_fcmp(
+; CHECK-SAME: double [[ARG0:%.*]]) {
+; CHECK-NEXT: [[V0:%.*]] = fptrunc double [[ARG0]] to float
+; CHECK-NEXT: [[V1:%.*]] = fcmp uno float [[V0]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[V1]]
+;
+ %v0 = fptrunc double %arg0 to float
+ %v1 = fcmp uno float %v0, 0.000000e+00
+ ret i1 %v1
+}
+
+define i1 @fptrunc_uno_fcmp_commuted(double %arg0) {
+; CHECK-LABEL: define i1 @fptrunc_uno_fcmp_commuted(
+; CHECK-SAME: double [[ARG0:%.*]]) {
+; CHECK-NEXT: [[V0:%.*]] = fptrunc double [[ARG0]] to float
+; CHECK-NEXT: [[V1:%.*]] = fcmp uno float [[V0]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[V1]]
+;
+ %v0 = fptrunc double %arg0 to float
+ %v1 = fcmp uno float 7.000000e+00, %v0
+ ret i1 %v1
+}
+
+define i1 @fptrunc_ord_fcmp(double %arg0) {
+; CHECK-LABEL: define i1 @fptrunc_ord_fcmp(
+; CHECK-SAME: double [[ARG0:%.*]]) {
+; CHECK-NEXT: [[V0:%.*]] = fptrunc double [[ARG0]] to float
+; CHECK-NEXT: [[V1:%.*]] = fcmp ord float [[V0]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[V1]]
+;
+ %v0 = fptrunc double %arg0 to float
+ %v1 = fcmp ord float %v0, 0.000000e+00
+ ret i1 %v1
+}
+
+define i1 @fptrunc_ord_fcmp_commuted(double %arg0) {
+; CHECK-LABEL: define i1 @fptrunc_ord_fcmp_commuted(
+; CHECK-SAME: double [[ARG0:%.*]]) {
+; CHECK-NEXT: [[V0:%.*]] = fptrunc double [[ARG0]] to float
+; CHECK-NEXT: [[V1:%.*]] = fcmp ord float [[V0]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[V1]]
+;
+ %v0 = fptrunc double %arg0 to float
+ %v1 = fcmp ord float 7.000000e+00, %v0
+ ret i1 %v1
+}
|
wermos
changed the title
fcmp ord/uno (fptrunc X), C to fcmp ord/uno X, Cfcmp ord/uno (fptrunc X), C to fcmp ord/uno X, C
🐧 Linux x64 Test Results
✅ The build succeeded and all tests passed. |
🪟 Windows x64 Test Results
✅ The build succeeded and all tests passed. |
Contributor
Author
|
Closing this as #185844 was merged. |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Fixes #185698.
Alive2 proof: https://alive2.llvm.org/ce/z/682cYT