[llvm] [LoopInterchange] Hoist isCompuatableLoopNest() in the control flow (PR #124247)
Madhur Amilkanthwar via llvm-commits
llvm-commits at lists.llvm.org
Tue Jan 28 00:11:34 PST 2025
https://github.com/madhur13490 updated https://github.com/llvm/llvm-project/pull/124247
>From dc9cfca155d3673c0b0c3de697b3074c63e0e71a Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madhura at nvidia.com>
Date: Wed, 22 Jan 2025 00:50:29 -0800
Subject: [PATCH 1/3] [LoopInterchange] Hoist isCompuatableLoopNest() in the
control flow
The profiling of the LLVM Test-suite reveals that a significant portion,
specifically 14,090 out of 139,323, loop nests were identified as non-viable
candidates for transformation, leading to the transform exiting from isComputableLoopNest()
without any action.
More importantly, dependence information was computed for these loop nests before
reaching the function isComputableLoopNest(), which does not require
DI and relies solely on scalar evolution (SE).
To enhance compile-time efficiency, this patch moves the call to isComputableLoopNest()
earlier in the control-flow, thereby avoiding unnecessary dependence calculations.
The impact of this change is evident on the compile-time-tracker, with the
overall geometric mean improvement recorded at 0.11%, while the lencode benchmark
gets a more substantial benefit of 0.44%.
This improvement can be tracked in the isc-ln-exp-2 branch under my repo.
---
.../lib/Transforms/Scalar/LoopInterchange.cpp | 52 ++++++++++---------
1 file changed, 28 insertions(+), 24 deletions(-)
diff --git a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
index 38fc682698c53e..c8d6ff638703eb 100644
--- a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
@@ -271,6 +271,26 @@ static bool hasSupportedLoopDepth(SmallVectorImpl<Loop *> &LoopList,
}
return true;
}
+
+static bool isComputableLoopNest(ScalarEvolution *SE, ArrayRef<Loop *> LoopList) {
+ for (Loop *L : LoopList) {
+ const SCEV *ExitCountOuter = SE->getBackedgeTakenCount(L);
+ if (isa<SCEVCouldNotCompute>(ExitCountOuter)) {
+ LLVM_DEBUG(dbgs() << "Couldn't compute backedge count\n");
+ return false;
+ }
+ if (L->getNumBackEdges() != 1) {
+ LLVM_DEBUG(dbgs() << "NumBackEdges is not equal to 1\n");
+ return false;
+ }
+ if (!L->getExitingBlock()) {
+ LLVM_DEBUG(dbgs() << "Loop doesn't have unique exit block\n");
+ return false;
+ }
+ }
+ return true;
+}
+
namespace {
/// LoopInterchangeLegality checks if it is legal to interchange the loop.
@@ -426,25 +446,6 @@ struct LoopInterchange {
return processLoopList(LoopList);
}
- bool isComputableLoopNest(ArrayRef<Loop *> LoopList) {
- for (Loop *L : LoopList) {
- const SCEV *ExitCountOuter = SE->getBackedgeTakenCount(L);
- if (isa<SCEVCouldNotCompute>(ExitCountOuter)) {
- LLVM_DEBUG(dbgs() << "Couldn't compute backedge count\n");
- return false;
- }
- if (L->getNumBackEdges() != 1) {
- LLVM_DEBUG(dbgs() << "NumBackEdges is not equal to 1\n");
- return false;
- }
- if (!L->getExitingBlock()) {
- LLVM_DEBUG(dbgs() << "Loop doesn't have unique exit block\n");
- return false;
- }
- }
- return true;
- }
-
unsigned selectLoopForInterchange(ArrayRef<Loop *> LoopList) {
// TODO: Add a better heuristic to select the loop to be interchanged based
// on the dependence matrix. Currently we select the innermost loop.
@@ -459,10 +460,6 @@ struct LoopInterchange {
"Unsupported depth of loop nest.");
unsigned LoopNestDepth = LoopList.size();
- if (!isComputableLoopNest(LoopList)) {
- LLVM_DEBUG(dbgs() << "Not valid loop candidate for interchange\n");
- return false;
- }
LLVM_DEBUG(dbgs() << "Processing LoopList of size = " << LoopNestDepth
<< "\n");
@@ -1756,10 +1753,17 @@ PreservedAnalyses LoopInterchangePass::run(LoopNest &LN,
// Ensure minimum depth of the loop nest to do the interchange.
if (!hasSupportedLoopDepth(LoopList, ORE))
return PreservedAnalyses::all();
+
+ // Ensure computable loop nest.
+ if (!isComputableLoopNest(&AR.SE, LoopList)) {
+ LLVM_DEBUG(dbgs() << "Not valid loop candidate for interchange\n");
+ return PreservedAnalyses::all();
+ }
+
DependenceInfo DI(&F, &AR.AA, &AR.SE, &AR.LI);
std::unique_ptr<CacheCost> CC =
CacheCost::getCacheCost(LN.getOutermostLoop(), AR, DI);
-
+
if (!LoopInterchange(&AR.SE, &AR.LI, &DI, &AR.DT, CC, &ORE).run(LN))
return PreservedAnalyses::all();
U.markLoopNestChanged(true);
>From ca4d8defd8e8b88867f83df6674039826d303752 Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madhura at nvidia.com>
Date: Mon, 27 Jan 2025 22:08:55 -0800
Subject: [PATCH 2/3] Update tests to account for new remarks
---
.../lib/Transforms/Scalar/LoopInterchange.cpp | 9 +-
.../loop-interchange-optimization-remarks.ll | 48 ++++++++
.../LoopInterchange/no-dependence-info.ll | 107 ++++++++++++++++++
.../pr43326-ideal-access-pattern.ll | 8 ++
.../Transforms/LoopInterchange/pr43326.ll | 8 ++
.../Transforms/LoopInterchange/pr48212.ll | 8 ++
.../reductions-across-inner-and-outer-loop.ll | 56 +++++++++
7 files changed, 243 insertions(+), 1 deletion(-)
create mode 100644 llvm/test/Transforms/LoopInterchange/no-dependence-info.ll
diff --git a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
index c8d6ff638703eb..611f0291926e34 100644
--- a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
@@ -272,7 +272,8 @@ static bool hasSupportedLoopDepth(SmallVectorImpl<Loop *> &LoopList,
return true;
}
-static bool isComputableLoopNest(ScalarEvolution *SE, ArrayRef<Loop *> LoopList) {
+static bool isComputableLoopNest(ScalarEvolution *SE,
+ ArrayRef<Loop *> LoopList) {
for (Loop *L : LoopList) {
const SCEV *ExitCountOuter = SE->getBackedgeTakenCount(L);
if (isa<SCEVCouldNotCompute>(ExitCountOuter)) {
@@ -1760,6 +1761,12 @@ PreservedAnalyses LoopInterchangePass::run(LoopNest &LN,
return PreservedAnalyses::all();
}
+ ORE.emit([&]() {
+ return OptimizationRemark(DEBUG_TYPE, "Dependence",
+ LN.getOutermostLoop().getStartLoc(), LN.getOutermostLoop().getHeader())
+ << "Computed dependence info, invoking the transform.";
+ });
+
DependenceInfo DI(&F, &AR.AA, &AR.SE, &AR.LI);
std::unique_ptr<CacheCost> CC =
CacheCost::getCacheCost(LN.getOutermostLoop(), AR, DI);
diff --git a/llvm/test/Transforms/LoopInterchange/loop-interchange-optimization-remarks.ll b/llvm/test/Transforms/LoopInterchange/loop-interchange-optimization-remarks.ll
index 3c7828a49477c3..ded5c344af7b78 100644
--- a/llvm/test/Transforms/LoopInterchange/loop-interchange-optimization-remarks.ll
+++ b/llvm/test/Transforms/LoopInterchange/loop-interchange-optimization-remarks.ll
@@ -58,6 +58,14 @@ for.end19:
ret void
}
+; CHECK: --- !Passed
+; CHECK-NEXT: Pass: loop-interchange
+; CHECK-NEXT: Name: Dependence
+; CHECK-NEXT: Function: test01
+; CHECK-NEXT: Args:
+; CHECK-NEXT: - String: Computed dependence info, invoking the transform.
+; CHECK-NEXT: ...
+
; CHECK: --- !Missed
; CHECK-NEXT: Pass: loop-interchange
; CHECK-NEXT: Name: Dependence
@@ -66,6 +74,14 @@ for.end19:
; CHECK-NEXT: - String: Cannot interchange loops due to dependences.
; CHECK-NEXT: ...
+; DELIN: --- !Passed
+; DELIN-NEXT: Pass: loop-interchange
+; DELIN-NEXT: Name: Dependence
+; DELIN-NEXT: Function: test01
+; DELIN-NEXT: Args:
+; DELIN-NEXT: - String: Computed dependence info, invoking the transform.
+; DELIN-NEXT: ...
+
; DELIN: --- !Missed
; DELIN-NEXT: Pass: loop-interchange
; DELIN-NEXT: Name: InterchangeNotProfitable
@@ -118,6 +134,14 @@ define void @test02(i32 %k, i32 %N) {
ret void
}
+; CHECK: --- !Passed
+; CHECK-NEXT: Pass: loop-interchange
+; CHECK-NEXT: Name: Dependence
+; CHECK-NEXT: Function: test02
+; CHECK-NEXT: Args:
+; CHECK-NEXT: - String: Computed dependence info, invoking the transform.
+; CHECK-NEXT: ...
+
; CHECK: --- !Missed
; CHECK-NEXT: Pass: loop-interchange
; CHECK-NEXT: Name: Dependence
@@ -126,6 +150,14 @@ define void @test02(i32 %k, i32 %N) {
; CHECK-NEXT: - String: Cannot interchange loops due to dependences.
; CHECK-NEXT: ...
+; DELIN: --- !Passed
+; DELIN-NEXT: Pass: loop-interchange
+; DELIN-NEXT: Name: Dependence
+; DELIN-NEXT: Function: test02
+; DELIN-NEXT: Args:
+; DELIN-NEXT: - String: Computed dependence info, invoking the transform.
+; DELIN-NEXT: ...
+
; DELIN: --- !Passed
; DELIN-NEXT: Pass: loop-interchange
; DELIN-NEXT: Name: Interchanged
@@ -174,6 +206,14 @@ for.body4: ; preds = %for.body4, %for.con
br i1 %exitcond, label %for.body4, label %for.cond.loopexit
}
+; CHECK: --- !Passed
+; CHECK-NEXT: Pass: loop-interchange
+; CHECK-NEXT: Name: Dependence
+; CHECK-NEXT: Function: test03
+; CHECK-NEXT: Args:
+; CHECK-NEXT: - String: Computed dependence info, invoking the transform.
+; CHECK-NEXT: ...
+
; CHECK: --- !Passed
; CHECK-NEXT: Pass: loop-interchange
; CHECK-NEXT: Name: Interchanged
@@ -182,6 +222,14 @@ for.body4: ; preds = %for.body4, %for.con
; CHECK-NEXT: - String: Loop interchanged with enclosing loop.
; CHECK-NEXT: ...
+; DELIN: --- !Passed
+; DELIN-NEXT: Pass: loop-interchange
+; DELIN-NEXT: Name: Dependence
+; DELIN-NEXT: Function: test03
+; DELIN-NEXT: Args:
+; DELIN-NEXT: - String: Computed dependence info, invoking the transform.
+; DELIN-NEXT: ...
+
; DELIN: --- !Passed
; DELIN-NEXT: Pass: loop-interchange
; DELIN-NEXT: Name: Interchanged
diff --git a/llvm/test/Transforms/LoopInterchange/no-dependence-info.ll b/llvm/test/Transforms/LoopInterchange/no-dependence-info.ll
new file mode 100644
index 00000000000000..cde9deeccbb83a
--- /dev/null
+++ b/llvm/test/Transforms/LoopInterchange/no-dependence-info.ll
@@ -0,0 +1,107 @@
+; RUN: opt %s -passes='loop-interchange' -pass-remarks=loop-interchange -disable-output 2>&1 | FileCheck --allow-empty %s
+
+target triple = "aarch64-unknown-linux-gnu"
+
+; CHECK-NOT: Computed dependence info, invoking the transform.
+
+define dso_local void @_foo(ptr noundef %a, ptr noundef %neg, ptr noundef %pos) {
+entry:
+ %a.addr = alloca ptr, align 8
+ %neg.addr = alloca ptr, align 8
+ %pos.addr = alloca ptr, align 8
+ %p = alloca i32, align 4
+ %q = alloca i32, align 4
+ %i = alloca i32, align 4
+ %cleanup.dest.slot = alloca i32, align 4
+ %j = alloca i32, align 4
+ store ptr %a, ptr %a.addr, align 8
+ store ptr %neg, ptr %neg.addr, align 8
+ store ptr %pos, ptr %pos.addr, align 8
+ store i32 0, ptr %p, align 4
+ store i32 0, ptr %q, align 4
+ store i32 0, ptr %i, align 4
+ br label %for.cond
+
+for.cond: ; preds = %for.inc16, %entry
+ %0 = load i32, ptr %i, align 4
+ %cmp = icmp slt i32 %0, 32
+ br i1 %cmp, label %for.body, label %for.cond.cleanup
+
+for.cond.cleanup: ; preds = %for.cond
+ store i32 2, ptr %cleanup.dest.slot, align 4
+ br label %for.end18
+
+for.body: ; preds = %for.cond
+ store i32 0, ptr %j, align 4
+ br label %for.cond1
+
+for.cond1: ; preds = %for.inc, %for.body
+ %1 = load i32, ptr %j, align 4
+ %cmp2 = icmp slt i32 %1, 32
+ br i1 %cmp2, label %for.body4, label %for.cond.cleanup3
+
+for.cond.cleanup3: ; preds = %for.cond1
+ store i32 5, ptr %cleanup.dest.slot, align 4
+ br label %for.end
+
+for.body4: ; preds = %for.cond1
+ %2 = load ptr, ptr %a.addr, align 8
+ %3 = load i32, ptr %i, align 4
+ %idxprom = sext i32 %3 to i64
+ %arrayidx = getelementptr inbounds i32, ptr %2, i64 %idxprom
+ %4 = load i32, ptr %arrayidx, align 4
+ %cmp5 = icmp slt i32 %4, 0
+ br i1 %cmp5, label %if.then, label %if.else
+
+if.then: ; preds = %for.body4
+ %5 = load ptr, ptr %a.addr, align 8
+ %6 = load i32, ptr %i, align 4
+ %idxprom6 = sext i32 %6 to i64
+ %arrayidx7 = getelementptr inbounds i32, ptr %5, i64 %idxprom6
+ %7 = load i32, ptr %arrayidx7, align 4
+ %8 = load ptr, ptr %neg.addr, align 8
+ %9 = load i32, ptr %p, align 4
+ %inc = add nsw i32 %9, 1
+ store i32 %inc, ptr %p, align 4
+ %idxprom8 = sext i32 %9 to i64
+ %arrayidx9 = getelementptr inbounds i32, ptr %8, i64 %idxprom8
+ store i32 %7, ptr %arrayidx9, align 4
+ br label %if.end
+
+if.else: ; preds = %for.body4
+ %10 = load ptr, ptr %a.addr, align 8
+ %11 = load i32, ptr %i, align 4
+ %idxprom10 = sext i32 %11 to i64
+ %arrayidx11 = getelementptr inbounds i32, ptr %10, i64 %idxprom10
+ %12 = load i32, ptr %arrayidx11, align 4
+ %13 = load ptr, ptr %pos.addr, align 8
+ %14 = load i32, ptr %q, align 4
+ %inc12 = add nsw i32 %14, 1
+ store i32 %inc12, ptr %q, align 4
+ %idxprom13 = sext i32 %14 to i64
+ %arrayidx14 = getelementptr inbounds i32, ptr %13, i64 %idxprom13
+ store i32 %12, ptr %arrayidx14, align 4
+ br label %if.end
+
+if.end: ; preds = %if.else, %if.then
+ br label %for.inc
+
+for.inc: ; preds = %if.end
+ %15 = load i32, ptr %j, align 4
+ %inc15 = add nsw i32 %15, 1
+ store i32 %inc15, ptr %j, align 4
+ br label %for.cond1
+
+for.end: ; preds = %for.cond.cleanup3
+ br label %for.inc16
+
+for.inc16: ; preds = %for.end
+ %16 = load i32, ptr %i, align 4
+ %inc17 = add nsw i32 %16, 1
+ store i32 %inc17, ptr %i, align 4
+ br label %for.cond
+
+for.end18: ; preds = %for.cond.cleanup
+ ret void
+}
+
diff --git a/llvm/test/Transforms/LoopInterchange/pr43326-ideal-access-pattern.ll b/llvm/test/Transforms/LoopInterchange/pr43326-ideal-access-pattern.ll
index def68ca3cd07eb..644788b756222c 100644
--- a/llvm/test/Transforms/LoopInterchange/pr43326-ideal-access-pattern.ll
+++ b/llvm/test/Transforms/LoopInterchange/pr43326-ideal-access-pattern.ll
@@ -14,6 +14,14 @@
; }
; }
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: pr43326-triply-nested
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Passed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: Interchanged
diff --git a/llvm/test/Transforms/LoopInterchange/pr43326.ll b/llvm/test/Transforms/LoopInterchange/pr43326.ll
index 4dbb06780a8984..2381f182ef3c96 100644
--- a/llvm/test/Transforms/LoopInterchange/pr43326.ll
+++ b/llvm/test/Transforms/LoopInterchange/pr43326.ll
@@ -8,6 +8,14 @@
@d = global i32 0
@e = global [1 x [1 x i32]] zeroinitializer
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: pr43326
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Passed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: Interchanged
diff --git a/llvm/test/Transforms/LoopInterchange/pr48212.ll b/llvm/test/Transforms/LoopInterchange/pr48212.ll
index b580794ab7d36e..6bb7bfd8260b64 100644
--- a/llvm/test/Transforms/LoopInterchange/pr48212.ll
+++ b/llvm/test/Transforms/LoopInterchange/pr48212.ll
@@ -2,6 +2,14 @@
; RUN: -verify-dom-info -verify-loop-info -verify-loop-lcssa 2>&1
; RUN: FileCheck --input-file=%t --check-prefix=REMARKS %s
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: pr48212
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Passed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: Interchanged
diff --git a/llvm/test/Transforms/LoopInterchange/reductions-across-inner-and-outer-loop.ll b/llvm/test/Transforms/LoopInterchange/reductions-across-inner-and-outer-loop.ll
index eea0c2635d5959..68a84b17ca4f64 100644
--- a/llvm/test/Transforms/LoopInterchange/reductions-across-inner-and-outer-loop.ll
+++ b/llvm/test/Transforms/LoopInterchange/reductions-across-inner-and-outer-loop.ll
@@ -5,6 +5,14 @@
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: test1
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Passed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: Interchanged
@@ -77,6 +85,14 @@ for1.loopexit: ; preds = %for1.inc
; In this test case, the inner reduction PHI %inner does not involve the outer
; reduction PHI %sum.outer, do not interchange.
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: test2
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Missed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: UnsupportedPHIOuter
@@ -114,6 +130,14 @@ for1.loopexit: ; preds = %for1.inc
; Check that we do not interchange if there is an additional instruction
; between the outer and inner reduction PHIs.
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: test3
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Missed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: UnsupportedPHIOuter
@@ -151,6 +175,14 @@ for1.loopexit: ; preds = %for1.inc
}
; Check that we do not interchange if reduction is stored in an invariant address inside inner loop
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: test4
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Missed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: Dependence
@@ -190,6 +222,14 @@ for1.loopexit: ; preds = %for1.inc
; Check that we do not interchange or crash if the PHI in the outer loop gets a
; constant from the inner loop.
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: test_constant_inner_loop_res
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Missed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: UnsupportedPHIOuter
@@ -229,6 +269,14 @@ for1.loopexit: ; preds = %for1.inc
; Floating point reductions are interchanged if all the fp instructions
; involved allow reassociation.
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: test5
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Passed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: Interchanged
@@ -269,6 +317,14 @@ for.exit: ; preds = %outer.inc
; Floating point reductions are not interchanged if not all the fp instructions
; involved allow reassociation.
+; REMARKS: --- !Passed
+; REMARKS-NEXT: Pass: loop-interchange
+; REMARKS-NEXT: Name: Dependence
+; REMARKS-NEXT: Function: test6
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT: - String: Computed dependence info, invoking the transform.
+; REMARKS-NEXT: ...
+
; REMARKS: --- !Missed
; REMARKS-NEXT: Pass: loop-interchange
; REMARKS-NEXT: Name: UnsupportedPHIOuter
>From 62d08e831d47ce4a14a619820a8be6da72445623 Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madhura at nvidia.com>
Date: Tue, 28 Jan 2025 00:11:20 -0800
Subject: [PATCH 3/3] Address formatting issues
---
llvm/lib/Transforms/Scalar/LoopInterchange.cpp | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
index 611f0291926e34..874d0b56515973 100644
--- a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp
@@ -1754,16 +1754,16 @@ PreservedAnalyses LoopInterchangePass::run(LoopNest &LN,
// Ensure minimum depth of the loop nest to do the interchange.
if (!hasSupportedLoopDepth(LoopList, ORE))
return PreservedAnalyses::all();
-
// Ensure computable loop nest.
if (!isComputableLoopNest(&AR.SE, LoopList)) {
- LLVM_DEBUG(dbgs() << "Not valid loop candidate for interchange\n");
- return PreservedAnalyses::all();
+ LLVM_DEBUG(dbgs() << "Not valid loop candidate for interchange\n");
+ return PreservedAnalyses::all();
}
ORE.emit([&]() {
return OptimizationRemark(DEBUG_TYPE, "Dependence",
- LN.getOutermostLoop().getStartLoc(), LN.getOutermostLoop().getHeader())
+ LN.getOutermostLoop().getStartLoc(),
+ LN.getOutermostLoop().getHeader())
<< "Computed dependence info, invoking the transform.";
});
More information about the llvm-commits
mailing list