[llvm] [InstCombine] Fold fcmp ogt (x - y), 0 into fcmp ogt x, y #85245 (PR #85506)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Apr 25 01:37:49 PDT 2024
https://github.com/SahilPatidar updated https://github.com/llvm/llvm-project/pull/85506
>From 03825688eb222b75133400d75215c7882771b73f Mon Sep 17 00:00:00 2001
From: SahilPatidar <patidarsahil2001 at gmail.com>
Date: Fri, 15 Mar 2024 16:49:36 +0530
Subject: [PATCH 1/7] [InstCombine] Fold fcmp ogt (x - y), 0 into fcmp ogt x, y
#85245
---
.../InstCombine/InstCombineCompares.cpp | 7 +++
llvm/test/Transforms/InstCombine/fcmp.ll | 43 +++++++++++++++++++
2 files changed, 50 insertions(+)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index fdadf729fbd541..5b1ca99e155f88 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -8045,6 +8045,13 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
Constant *RHSC;
if (match(Op0, m_Instruction(LHSI)) && match(Op1, m_Constant(RHSC))) {
switch (LHSI->getOpcode()) {
+ case Instruction::FSub:
+ if ((Pred == FCmpInst::FCMP_OGT || Pred == FCmpInst::FCMP_OLT ||
+ Pred == FCmpInst::FCMP_ONE) &&
+ match(RHSC, m_AnyZeroFP()) &&
+ match(LHSI, m_FSub(m_Value(X), m_Value(Y))))
+ return new FCmpInst(Pred, X, Y);
+ break;
case Instruction::PHI:
if (Instruction *NV = foldOpIntoPhi(I, cast<PHINode>(LHSI)))
return NV;
diff --git a/llvm/test/Transforms/InstCombine/fcmp.ll b/llvm/test/Transforms/InstCombine/fcmp.ll
index 389264e2f70759..aa374a6586ad76 100644
--- a/llvm/test/Transforms/InstCombine/fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/fcmp.ll
@@ -1486,3 +1486,46 @@ define i1 @fcmp_fadd_fast_zero(float %x, float %y) {
%cmp = fcmp ugt float %add, %y
ret i1 %cmp
}
+
+define i1 @fcmp_ogt_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_ogt_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp ogt float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_olt_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_olt_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp olt float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_one_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_one_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp one float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp one float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_fsub_neg_zero(float %x, float %y) {
+; CHECK-LABEL: @fcmp_fsub_neg_zero(
+; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: call void @use(float [[FS]])
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt float [[X]], [[Y]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ call void @use(float %fs)
+ %cmp = fcmp ogt float %fs, -0.000000e+00
+ ret i1 %cmp
+}
>From 71f8f994bc22b8946cb037852f127f0ed2e87f61 Mon Sep 17 00:00:00 2001
From: SahilPatidar <patidarsahil2001 at gmail.com>
Date: Mon, 18 Mar 2024 15:11:30 +0530
Subject: [PATCH 2/7] Added tests to verify fast math flag behavior
---
llvm/test/Transforms/InstCombine/fcmp.ll | 97 +++++++++++++++++++++---
1 file changed, 87 insertions(+), 10 deletions(-)
diff --git a/llvm/test/Transforms/InstCombine/fcmp.ll b/llvm/test/Transforms/InstCombine/fcmp.ll
index aa374a6586ad76..cf7c04a34cb81d 100644
--- a/llvm/test/Transforms/InstCombine/fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/fcmp.ll
@@ -1517,15 +1517,92 @@ define i1 @fcmp_one_fsub_const(float %x, float %y) {
ret i1 %cmp
}
-define i1 @fcmp_fsub_neg_zero(float %x, float %y) {
-; CHECK-LABEL: @fcmp_fsub_neg_zero(
-; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: call void @use(float [[FS]])
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt float [[X]], [[Y]]
-; CHECK-NEXT: ret i1 [[CMP]]
+define <8 x i1> @fcmp_vec_ogt_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ogt_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub float %x, %y
- call void @use(float %fs)
- %cmp = fcmp ogt float %fs, -0.000000e+00
- ret i1 %cmp
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp ogt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_olt_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_olt_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp olt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_one_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_one_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp one <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp one <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_ogt_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ogt_fm_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp ogt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_olt_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_olt_fm_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp olt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_one_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_one_fm_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp one <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp one <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @ffcmp_vec_ogt_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @ffcmp_vec_ogt_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp fast ogt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @ffcmp_vec_olt_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @ffcmp_vec_olt_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp fast olt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @ffcmp_vec_one_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @ffcmp_vec_one_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp one <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp fast one <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
}
>From efba0d03d3d88d037eca1607697982c9aff9cf26 Mon Sep 17 00:00:00 2001
From: SahilPatidar <patidarsahil2001 at gmail.com>
Date: Fri, 22 Mar 2024 22:23:05 +0530
Subject: [PATCH 3/7] Modifies code and updates relevant tests
---
.../InstCombine/InstCombineCompares.cpp | 35 +++-
llvm/test/Transforms/InstCombine/fcmp.ll | 198 ++++++++++++++----
2 files changed, 192 insertions(+), 41 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 5b1ca99e155f88..3433afdec6a1d2 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -8046,11 +8046,36 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
if (match(Op0, m_Instruction(LHSI)) && match(Op1, m_Constant(RHSC))) {
switch (LHSI->getOpcode()) {
case Instruction::FSub:
- if ((Pred == FCmpInst::FCMP_OGT || Pred == FCmpInst::FCMP_OLT ||
- Pred == FCmpInst::FCMP_ONE) &&
- match(RHSC, m_AnyZeroFP()) &&
- match(LHSI, m_FSub(m_Value(X), m_Value(Y))))
- return new FCmpInst(Pred, X, Y);
+ switch (Pred) {
+ default:
+ break;
+ case FCmpInst::FCMP_UGT:
+ case FCmpInst::FCMP_ULT:
+ case FCmpInst::FCMP_UNE:
+ case FCmpInst::FCMP_OEQ:
+ case FCmpInst::FCMP_OGE:
+ case FCmpInst::FCMP_OLE: {
+ BinaryOperator *SubI = cast<BinaryOperator>(LHSI);
+ if (!computeKnownFPClass(SubI->getOperand(0), SubI->getFastMathFlags(),
+ fcInf, LHSI, 0)
+ .isKnownNeverInfinity() &&
+ !computeKnownFPClass(SubI->getOperand(1), SubI->getFastMathFlags(),
+ fcInf, LHSI, 0)
+ .isKnownNeverInfinity())
+ break;
+ }
+ LLVM_FALLTHROUGH;
+ case FCmpInst::FCMP_OGT:
+ case FCmpInst::FCMP_OLT:
+ case FCmpInst::FCMP_ONE:
+ case FCmpInst::FCMP_UEQ:
+ case FCmpInst::FCMP_UGE:
+ case FCmpInst::FCMP_ULE:
+ if (match(RHSC, m_AnyZeroFP()) &&
+ match(LHSI, m_FSub(m_Value(X), m_Value(Y))))
+ return new FCmpInst(Pred, X, Y);
+ break;
+ }
break;
case Instruction::PHI:
if (Instruction *NV = foldOpIntoPhi(I, cast<PHINode>(LHSI)))
diff --git a/llvm/test/Transforms/InstCombine/fcmp.ll b/llvm/test/Transforms/InstCombine/fcmp.ll
index cf7c04a34cb81d..f2431eccefed0f 100644
--- a/llvm/test/Transforms/InstCombine/fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/fcmp.ll
@@ -1517,38 +1517,164 @@ define i1 @fcmp_one_fsub_const(float %x, float %y) {
ret i1 %cmp
}
-define <8 x i1> @fcmp_vec_ogt_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ogt_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+define i1 @fcmp_oeq_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_oeq_fsub_const(
+; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp oeq float [[FS]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp oeq float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_oge_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_oge_fsub_const(
+; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp oge float [[FS]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp oge float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_ole_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_ole_fsub_const(
+; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ole float [[FS]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp ole float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_ueq_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_ueq_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ueq float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp ueq float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_uge_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_uge_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp uge float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp uge float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_ule_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_ule_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ule float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp ule float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_ugt_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_ugt_fsub_const(
+; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ugt float [[FS]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp ugt float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_ult_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_ult_fsub_const(
+; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ult float [[FS]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp ult float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define i1 @fcmp_une_fsub_const(float %x, float %y) {
+; CHECK-LABEL: @fcmp_une_fsub_const(
+; CHECK-NEXT: [[FS:%.*]] = fsub float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp une float [[FS]], 0.000000e+00
+; CHECK-NEXT: ret i1 [[CMP]]
+;
+ %fs = fsub float %x, %y
+ %cmp = fcmp une float %fs, 0.000000e+00
+ ret i1 %cmp
+}
+
+define <8 x i1> @fcmp_vec_uge_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_uge_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp uge <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub <8 x float> %x, %y
- %cmp = fcmp ogt <8 x float> %fs, zeroinitializer
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp uge <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_olt_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_olt_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ule_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ule_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ule <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub <8 x float> %x, %y
- %cmp = fcmp olt <8 x float> %fs, zeroinitializer
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp ule <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_one_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_one_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp one <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ueq_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ueq_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ueq <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub <8 x float> %x, %y
- %cmp = fcmp one <8 x float> %fs, zeroinitializer
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp ueq <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ogt_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ogt_fm_fsub_const(
+define <8 x i1> @fcmp_vec_oge_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_oge_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp oge <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp oge <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_ole_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ole_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ole <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp ole <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_oeq_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_oeq_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp oeq <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp oeq <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_vec_ogt_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ogt_fast_fsub_const(
; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
@@ -1557,8 +1683,8 @@ define <8 x i1> @fcmp_vec_ogt_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_olt_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_olt_fm_fsub_const(
+define <8 x i1> @fcmp_vec_olt_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_olt_fast_fsub_const(
; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
@@ -1567,8 +1693,8 @@ define <8 x i1> @fcmp_vec_olt_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_one_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_one_fm_fsub_const(
+define <8 x i1> @fcmp_vec_one_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_one_fast_fsub_const(
; CHECK-NEXT: [[CMP:%.*]] = fcmp one <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
@@ -1577,32 +1703,32 @@ define <8 x i1> @fcmp_vec_one_fm_fsub_const(<8 x float> %x, <8 x float> %y) {
ret <8 x i1> %cmp
}
-define <8 x i1> @ffcmp_vec_ogt_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @ffcmp_vec_ogt_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ugt_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ugt_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ugt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub <8 x float> %x, %y
- %cmp = fcmp fast ogt <8 x float> %fs, zeroinitializer
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp ugt <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @ffcmp_vec_olt_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @ffcmp_vec_olt_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ult_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ult_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ult <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub <8 x float> %x, %y
- %cmp = fcmp fast olt <8 x float> %fs, zeroinitializer
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp ult <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @ffcmp_vec_one_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @ffcmp_vec_one_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp one <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_une_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_une_fast_fsub_const(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp une <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub <8 x float> %x, %y
- %cmp = fcmp fast one <8 x float> %fs, zeroinitializer
+ %fs = fsub fast <8 x float> %x, %y
+ %cmp = fcmp une <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
>From c01fd1ad781b55aed50796c804cbd473f011f1e1 Mon Sep 17 00:00:00 2001
From: SahilPatidar <patidarsahil2001 at gmail.com>
Date: Wed, 27 Mar 2024 16:18:22 +0530
Subject: [PATCH 4/7] added fmf flag to fcmp
---
.../InstCombine/InstCombineCompares.cpp | 2 +-
llvm/test/Transforms/InstCombine/fcmp.ll | 96 +++++++++----------
2 files changed, 49 insertions(+), 49 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 3433afdec6a1d2..e0b65e17c04348 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -8073,7 +8073,7 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
case FCmpInst::FCMP_ULE:
if (match(RHSC, m_AnyZeroFP()) &&
match(LHSI, m_FSub(m_Value(X), m_Value(Y))))
- return new FCmpInst(Pred, X, Y);
+ return new FCmpInst(Pred, X, Y, "", &I);
break;
}
break;
diff --git a/llvm/test/Transforms/InstCombine/fcmp.ll b/llvm/test/Transforms/InstCombine/fcmp.ll
index f2431eccefed0f..f42e1d49a43abd 100644
--- a/llvm/test/Transforms/InstCombine/fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/fcmp.ll
@@ -1613,122 +1613,122 @@ define i1 @fcmp_une_fsub_const(float %x, float %y) {
ret i1 %cmp
}
-define <8 x i1> @fcmp_vec_uge_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_uge_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp uge <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_uge_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_uge_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast uge <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp uge <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast uge <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ule_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ule_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ule <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ule_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ule_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ule <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp ule <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast ule <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ueq_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ueq_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ueq <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ueq_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ueq_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ueq <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp ueq <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast ueq <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_oge_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_oge_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp oge <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_oge_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_oge_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast oge <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp oge <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast oge <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ole_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ole_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ole <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ole_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ole_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ole <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp ole <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast ole <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_oeq_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_oeq_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp oeq <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_oeq_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_oeq_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast oeq <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp oeq <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast oeq <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ogt_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ogt_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ogt_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ogt_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ogt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp ogt <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast ogt <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_olt_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_olt_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_olt_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_olt_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast olt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp olt <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast olt <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_one_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_one_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp one <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_one_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_one_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast one <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp one <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast one <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ugt_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ugt_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ugt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ugt_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ugt_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ugt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp ugt <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast ugt <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ult_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ult_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ult <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_ult_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_ult_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ult <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp ult <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast ult <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_une_fast_fsub_const(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_une_fast_fsub_const(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp une <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_vec_une_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_vec_une_fsub_const_fmf(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast une <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp une <8 x float> %fs, zeroinitializer
+ %cmp = fcmp fast une <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
>From 11c4c530197ee7b4da46db71720486dcba7c824a Mon Sep 17 00:00:00 2001
From: SahilPatidar <patidarsahil2001 at gmail.com>
Date: Sat, 13 Apr 2024 15:10:25 +0530
Subject: [PATCH 5/7] Fix code to handle ninf and nnan
---
.../InstCombine/InstCombineCompares.cpp | 8 +-
llvm/test/Transforms/InstCombine/fcmp.ll | 244 ++++++++++++++----
2 files changed, 190 insertions(+), 62 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index e0b65e17c04348..106d94370ef526 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -8056,12 +8056,8 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
case FCmpInst::FCMP_OGE:
case FCmpInst::FCMP_OLE: {
BinaryOperator *SubI = cast<BinaryOperator>(LHSI);
- if (!computeKnownFPClass(SubI->getOperand(0), SubI->getFastMathFlags(),
- fcInf, LHSI, 0)
- .isKnownNeverInfinity() &&
- !computeKnownFPClass(SubI->getOperand(1), SubI->getFastMathFlags(),
- fcInf, LHSI, 0)
- .isKnownNeverInfinity())
+ if (!isKnownNeverInfOrNaN(SubI->getOperand(0), 0, SQ) &&
+ !isKnownNeverInfOrNaN(SubI->getOperand(1), 0, SQ))
break;
}
LLVM_FALLTHROUGH;
diff --git a/llvm/test/Transforms/InstCombine/fcmp.ll b/llvm/test/Transforms/InstCombine/fcmp.ll
index f42e1d49a43abd..bbf7e4f80065eb 100644
--- a/llvm/test/Transforms/InstCombine/fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/fcmp.ll
@@ -1613,109 +1613,240 @@ define i1 @fcmp_une_fsub_const(float %x, float %y) {
ret i1 %cmp
}
-define <8 x i1> @fcmp_vec_uge_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_uge_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast uge <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_uge_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_uge_fsub_const_ninf_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf uge <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast uge <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf uge <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ule_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ule_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ule <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_ule_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ule_fsub_const_ninf_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ule <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast ule <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf ule <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ueq_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ueq_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ueq <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_ueq_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ueq_fsub_const_ninf_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ueq <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast ueq <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf ueq <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_oge_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_oge_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast oge <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_oge_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_oge_fsub_const_ninf_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf oge <8 x float> [[FS]], zeroinitializer
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast oge <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf oge <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ole_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ole_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ole <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_ole_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ole_fsub_const_ninf_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ole <8 x float> [[FS]], zeroinitializer
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast ole <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf ole <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_oeq_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_oeq_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast oeq <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_oeq_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_oeq_fsub_const_ninf_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf oeq <8 x float> [[FS]], zeroinitializer
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast oeq <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf oeq <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ogt_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ogt_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_ogt_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ogt_fsub_const_ninf_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ogt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast ogt <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf ogt <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_olt_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_olt_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast olt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_olt_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_olt_fsub_const_ninf_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf olt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast olt <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf olt <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_one_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_one_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast one <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_one_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_one_fsub_const_ninf_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf one <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast one <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf one <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ugt_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ugt_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ugt <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_ugt_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ugt_fsub_const_ninf_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ugt <8 x float> [[FS]], zeroinitializer
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast ugt <8 x float> %fs, zeroinitializer
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf ugt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ult_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ult_fsub_const_ninf_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ult <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf ult <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_une_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_une_fsub_const_ninf_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf une <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub ninf <8 x float> %x, %y
+ %cmp = fcmp ninf une <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_uge_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_uge_fsub_const_nnan_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan uge <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan uge <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ule_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ule_fsub_const_nnan_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ule <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan ule <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ueq_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ueq_fsub_const_nnan_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ueq <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan ueq <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_oge_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_oge_fsub_const_nnan_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan oge <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan oge <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ole_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ole_fsub_const_nnan_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ole <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan ole <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_oeq_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_oeq_fsub_const_nnan_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan oeq <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan oeq <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ogt_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ogt_fsub_const_nnan_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ogt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan ogt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_olt_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_olt_fsub_const_nnan_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan olt <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan olt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_one_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_one_fsub_const_nnan_vec(
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan one <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan one <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ugt_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ugt_fsub_const_nnan_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ugt <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan ugt <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_ult_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_ult_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ult <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_ult_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_ult_fsub_const_nnan_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub fast <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ult <8 x float> [[FS]], zeroinitializer
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
@@ -1723,9 +1854,10 @@ define <8 x i1> @fcmp_vec_ult_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
ret <8 x i1> %cmp
}
-define <8 x i1> @fcmp_vec_une_fsub_const_fmf(<8 x float> %x, <8 x float> %y) {
-; CHECK-LABEL: @fcmp_vec_une_fsub_const_fmf(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast une <8 x float> [[X:%.*]], [[Y:%.*]]
+define <8 x i1> @fcmp_une_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
+; CHECK-LABEL: @fcmp_une_fsub_const_nnan_vec(
+; CHECK-NEXT: [[FS:%.*]] = fsub fast <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp fast une <8 x float> [[FS]], zeroinitializer
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub fast <8 x float> %x, %y
>From f2b995c1f3ba6ad8a62df298a615be8e1099271d Mon Sep 17 00:00:00 2001
From: SahilPatidar <patidarsahil2001 at gmail.com>
Date: Wed, 17 Apr 2024 15:06:58 +0530
Subject: [PATCH 6/7] fix: Handle ninf values (update test)
---
.../InstCombine/InstCombineCompares.cpp | 15 ++++---
llvm/test/Transforms/InstCombine/fcmp.ll | 44 +++++++------------
2 files changed, 26 insertions(+), 33 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 106d94370ef526..2fe35cb2a6ee8f 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -8055,9 +8055,11 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
case FCmpInst::FCMP_OEQ:
case FCmpInst::FCMP_OGE:
case FCmpInst::FCMP_OLE: {
- BinaryOperator *SubI = cast<BinaryOperator>(LHSI);
- if (!isKnownNeverInfOrNaN(SubI->getOperand(0), 0, SQ) &&
- !isKnownNeverInfOrNaN(SubI->getOperand(1), 0, SQ))
+ if (!LHSI->hasNoNaNs() && !LHSI->hasNoInfs() &&
+ !isKnownNeverInfinity(LHSI->getOperand(0), /*Depth=*/0,
+ getSimplifyQuery().getWithInstruction(&I)) &&
+ !isKnownNeverInfinity(LHSI->getOperand(1), /*Depth=*/0,
+ getSimplifyQuery().getWithInstruction(&I)))
break;
}
LLVM_FALLTHROUGH;
@@ -8068,8 +8070,11 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
case FCmpInst::FCMP_UGE:
case FCmpInst::FCMP_ULE:
if (match(RHSC, m_AnyZeroFP()) &&
- match(LHSI, m_FSub(m_Value(X), m_Value(Y))))
- return new FCmpInst(Pred, X, Y, "", &I);
+ match(LHSI, m_FSub(m_Value(X), m_Value(Y)))) {
+ replaceOperand(I, 0, X);
+ replaceOperand(I, 1, Y);
+ return &I;
+ }
break;
}
break;
diff --git a/llvm/test/Transforms/InstCombine/fcmp.ll b/llvm/test/Transforms/InstCombine/fcmp.ll
index bbf7e4f80065eb..d20bf609973426 100644
--- a/llvm/test/Transforms/InstCombine/fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/fcmp.ll
@@ -1645,8 +1645,7 @@ define <8 x i1> @fcmp_ueq_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_oge_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_oge_fsub_const_ninf_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf oge <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf oge <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub ninf <8 x float> %x, %y
@@ -1656,8 +1655,7 @@ define <8 x i1> @fcmp_oge_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_ole_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_ole_fsub_const_ninf_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ole <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ole <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub ninf <8 x float> %x, %y
@@ -1667,8 +1665,7 @@ define <8 x i1> @fcmp_ole_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_oeq_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_oeq_fsub_const_ninf_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf oeq <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf oeq <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub ninf <8 x float> %x, %y
@@ -1708,8 +1705,7 @@ define <8 x i1> @fcmp_one_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_ugt_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_ugt_fsub_const_ninf_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ugt <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ugt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub ninf <8 x float> %x, %y
@@ -1719,8 +1715,7 @@ define <8 x i1> @fcmp_ugt_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_ult_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_ult_fsub_const_ninf_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ult <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf ult <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub ninf <8 x float> %x, %y
@@ -1730,8 +1725,7 @@ define <8 x i1> @fcmp_ult_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_une_fsub_const_ninf_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_une_fsub_const_ninf_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub ninf <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf une <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ninf une <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub ninf <8 x float> %x, %y
@@ -1771,8 +1765,7 @@ define <8 x i1> @fcmp_ueq_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_oge_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_oge_fsub_const_nnan_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan oge <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan oge <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub nnan <8 x float> %x, %y
@@ -1782,8 +1775,7 @@ define <8 x i1> @fcmp_oge_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_ole_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_ole_fsub_const_nnan_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ole <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ole <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub nnan <8 x float> %x, %y
@@ -1793,8 +1785,7 @@ define <8 x i1> @fcmp_ole_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_oeq_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_oeq_fsub_const_nnan_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan oeq <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan oeq <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub nnan <8 x float> %x, %y
@@ -1834,8 +1825,7 @@ define <8 x i1> @fcmp_one_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_ugt_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_ugt_fsub_const_nnan_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub nnan <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ugt <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ugt <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
%fs = fsub nnan <8 x float> %x, %y
@@ -1845,22 +1835,20 @@ define <8 x i1> @fcmp_ugt_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
define <8 x i1> @fcmp_ult_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_ult_fsub_const_nnan_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub fast <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast ult <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan ult <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast ult <8 x float> %fs, zeroinitializer
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan ult <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
define <8 x i1> @fcmp_une_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
; CHECK-LABEL: @fcmp_une_fsub_const_nnan_vec(
-; CHECK-NEXT: [[FS:%.*]] = fsub fast <8 x float> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = fcmp fast une <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: [[CMP:%.*]] = fcmp nnan une <8 x float> [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret <8 x i1> [[CMP]]
;
- %fs = fsub fast <8 x float> %x, %y
- %cmp = fcmp fast une <8 x float> %fs, zeroinitializer
+ %fs = fsub nnan <8 x float> %x, %y
+ %cmp = fcmp nnan une <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
>From 773d159ac914c42424ae5d8e3fb0353419bc5a04 Mon Sep 17 00:00:00 2001
From: SahilPatidar <patidarsahil2001 at gmail.com>
Date: Thu, 25 Apr 2024 11:47:03 +0530
Subject: [PATCH 7/7] adds test and update to check denormals being flushed
---
.../InstCombine/InstCombineCompares.cpp | 9 ++-
llvm/test/Transforms/InstCombine/fcmp.ll | 61 ++++++++++++++-----
2 files changed, 53 insertions(+), 17 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 2fe35cb2a6ee8f..33e7e948aa11c0 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -8056,9 +8056,9 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
case FCmpInst::FCMP_OGE:
case FCmpInst::FCMP_OLE: {
if (!LHSI->hasNoNaNs() && !LHSI->hasNoInfs() &&
- !isKnownNeverInfinity(LHSI->getOperand(0), /*Depth=*/0,
- getSimplifyQuery().getWithInstruction(&I)) &&
!isKnownNeverInfinity(LHSI->getOperand(1), /*Depth=*/0,
+ getSimplifyQuery().getWithInstruction(&I)) &&
+ !isKnownNeverInfinity(LHSI->getOperand(0), /*Depth=*/0,
getSimplifyQuery().getWithInstruction(&I)))
break;
}
@@ -8070,7 +8070,10 @@ Instruction *InstCombinerImpl::visitFCmpInst(FCmpInst &I) {
case FCmpInst::FCMP_UGE:
case FCmpInst::FCMP_ULE:
if (match(RHSC, m_AnyZeroFP()) &&
- match(LHSI, m_FSub(m_Value(X), m_Value(Y)))) {
+ match(LHSI, m_FSub(m_Value(X), m_Value(Y))) &&
+ I.getFunction()->getDenormalMode(
+ LHSI->getType()->getScalarType()->getFltSemantics()) ==
+ DenormalMode::getIEEE()) {
replaceOperand(I, 0, X);
replaceOperand(I, 1, Y);
return &I;
diff --git a/llvm/test/Transforms/InstCombine/fcmp.ll b/llvm/test/Transforms/InstCombine/fcmp.ll
index d20bf609973426..a57b49e4628597 100644
--- a/llvm/test/Transforms/InstCombine/fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/fcmp.ll
@@ -1289,7 +1289,7 @@ define <1 x i1> @bitcast_1vec_eq0(i32 %x) {
define i1 @fcmp_fadd_zero_ugt(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_ugt(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ugt float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ugt float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1299,7 +1299,7 @@ define i1 @fcmp_fadd_zero_ugt(float %x, float %y) {
define i1 @fcmp_fadd_zero_uge(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_uge(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp uge float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp uge float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1309,7 +1309,7 @@ define i1 @fcmp_fadd_zero_uge(float %x, float %y) {
define i1 @fcmp_fadd_zero_ogt(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_ogt(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1319,7 +1319,7 @@ define i1 @fcmp_fadd_zero_ogt(float %x, float %y) {
define i1 @fcmp_fadd_zero_oge(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_oge(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp oge float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp oge float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1329,7 +1329,7 @@ define i1 @fcmp_fadd_zero_oge(float %x, float %y) {
define i1 @fcmp_fadd_zero_ult(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_ult(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ult float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ult float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1339,7 +1339,7 @@ define i1 @fcmp_fadd_zero_ult(float %x, float %y) {
define i1 @fcmp_fadd_zero_ule(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_ule(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ule float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ule float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1349,7 +1349,7 @@ define i1 @fcmp_fadd_zero_ule(float %x, float %y) {
define i1 @fcmp_fadd_zero_olt(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_olt(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp olt float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1359,7 +1359,7 @@ define i1 @fcmp_fadd_zero_olt(float %x, float %y) {
define i1 @fcmp_fadd_zero_ole(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_ole(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ole float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ole float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1369,7 +1369,7 @@ define i1 @fcmp_fadd_zero_ole(float %x, float %y) {
define i1 @fcmp_fadd_zero_oeq(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_oeq(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp oeq float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp oeq float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1379,7 +1379,7 @@ define i1 @fcmp_fadd_zero_oeq(float %x, float %y) {
define i1 @fcmp_fadd_zero_one(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_one(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp one float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp one float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1389,7 +1389,7 @@ define i1 @fcmp_fadd_zero_one(float %x, float %y) {
define i1 @fcmp_fadd_zero_ueq(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_ueq(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ueq float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ueq float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1399,7 +1399,7 @@ define i1 @fcmp_fadd_zero_ueq(float %x, float %y) {
define i1 @fcmp_fadd_zero_une(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_une(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp une float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp une float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1409,7 +1409,7 @@ define i1 @fcmp_fadd_zero_une(float %x, float %y) {
define i1 @fcmp_fadd_zero_ord(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_ord(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp ord float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ord float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1419,7 +1419,7 @@ define i1 @fcmp_fadd_zero_ord(float %x, float %y) {
define i1 @fcmp_fadd_zero_uno(float %x, float %y) {
; CHECK-LABEL: @fcmp_fadd_zero_uno(
-; CHECK-NEXT: [[CMP:%.*]] = fcmp uno float [[ADD:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp uno float [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%add = fadd float %x, 0.000000e+00
@@ -1852,3 +1852,36 @@ define <8 x i1> @fcmp_une_fsub_const_nnan_vec(<8 x float> %x, <8 x float> %y) {
%cmp = fcmp nnan une <8 x float> %fs, zeroinitializer
ret <8 x i1> %cmp
}
+
+define <8 x i1> @fcmp_ugt_fsub_const_vec_denormal_positive-zero(<8 x float> %x, <8 x float> %y) "denormal-fp-math"="positive-zero" {
+; CHECK-LABEL: @fcmp_ugt_fsub_const_vec_denormal_positive-zero(
+; CHECK-NEXT: [[FS:%.*]] = fsub <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp ogt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ogt_fsub_const_vec_denormal_dynamic(<8 x float> %x, <8 x float> %y) "denormal-fp-math"="dynamic,dynamic" {
+; CHECK-LABEL: @fcmp_ogt_fsub_const_vec_denormal_dynamic(
+; CHECK-NEXT: [[FS:%.*]] = fsub <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp ogt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
+
+define <8 x i1> @fcmp_ogt_fsub_const_vec_denormal_preserve-sign(<8 x float> %x, <8 x float> %y) "denormal-fp-math"="preserve-sign" {
+; CHECK-LABEL: @fcmp_ogt_fsub_const_vec_denormal_preserve-sign(
+; CHECK-NEXT: [[FS:%.*]] = fsub <8 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[FS]], zeroinitializer
+; CHECK-NEXT: ret <8 x i1> [[CMP]]
+;
+ %fs = fsub <8 x float> %x, %y
+ %cmp = fcmp ogt <8 x float> %fs, zeroinitializer
+ ret <8 x i1> %cmp
+}
More information about the llvm-commits
mailing list