https://github.com/madhur13490 updated
https://github.com/llvm/llvm-project/pull/142686
>From e2f53441f1a2d35af7c0d6d177afea79c7af2c47 Mon Sep 17 00:00:00 2001
From: Sebastian Pop <s...@nvidia.com>
Date: Thu, 22 May 2025 13:50:38 +0000
Subject: [PATCH 01/10] add -floop-fuse to clang and flang
---
clang/include/clang/Basic/CodeGenOptions.def | 1 +
clang/include/clang/Driver/Options.td | 4 ++++
clang/lib/CodeGen/BackendUtil.cpp | 2 ++
clang/lib/Driver/ToolChains/Clang.cpp | 1 +
clang/lib/Driver/ToolChains/CommonArgs.cpp | 10 ++++++++--
clang/lib/Driver/ToolChains/Flang.cpp | 2 ++
clang/lib/Frontend/CompilerInvocation.cpp | 6 ++++++
clang/test/Driver/clang_f_opts.c | 7 +++++++
flang/docs/ReleaseNotes.md | 4 ++++
flang/include/flang/Frontend/CodeGenOptions.def | 1 +
flang/lib/Frontend/CompilerInvocation.cpp | 3 +++
flang/lib/Frontend/FrontendActions.cpp | 1 +
flang/test/Driver/loop-fuse.f90 | 17 +++++++++++++++++
llvm/include/llvm/Passes/PassBuilder.h | 3 +++
llvm/lib/Passes/PassBuilderPipelines.cpp | 13 ++++++++++++-
15 files changed, 72 insertions(+), 3 deletions(-)
create mode 100644 flang/test/Driver/loop-fuse.f90
diff --git a/clang/include/clang/Basic/CodeGenOptions.def
b/clang/include/clang/Basic/CodeGenOptions.def
index fda0da99b60c0..4c4f0970dfff0 100644
--- a/clang/include/clang/Basic/CodeGenOptions.def
+++ b/clang/include/clang/Basic/CodeGenOptions.def
@@ -322,6 +322,7 @@ CODEGENOPT(TimeTrace , 1, 0, Benign) ///< Set when
-ftime-trace is enabl
VALUE_CODEGENOPT(TimeTraceGranularity, 32, 500, Benign) ///< Minimum time
granularity (in microseconds),
///< traced by time
profiler
CODEGENOPT(InterchangeLoops , 1, 0, Benign) ///< Run loop-interchange.
+CODEGENOPT(FuseLoops , 1, 0, Benign) ///< Run loop-fuse.
CODEGENOPT(UnrollLoops , 1, 0, Benign) ///< Control whether loops are
unrolled.
CODEGENOPT(RerollLoops , 1, 0, Benign) ///< Control whether loops are
rerolled.
CODEGENOPT(NoUseJumpTables , 1, 0, Benign) ///< Set when -fno-jump-tables is
enabled.
diff --git a/clang/include/clang/Driver/Options.td
b/clang/include/clang/Driver/Options.td
index a7c514e809aa9..13a8502c6c7b5 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -4304,6 +4304,10 @@ def floop_interchange : Flag<["-"],
"floop-interchange">, Group<f_Group>,
HelpText<"Enable the loop interchange pass">, Visibility<[ClangOption,
CC1Option, FlangOption, FC1Option]>;
def fno_loop_interchange: Flag<["-"], "fno-loop-interchange">, Group<f_Group>,
HelpText<"Disable the loop interchange pass">, Visibility<[ClangOption,
CC1Option, FlangOption, FC1Option]>;
+def floop_fuse : Flag<["-"], "floop-fuse">, Group<f_Group>,
+ HelpText<"Enable the loop fuse pass">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
+def fno_loop_fuse: Flag<["-"], "fno-loop-fuse">, Group<f_Group>,
+ HelpText<"Disable the loop fuse pass">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
def funroll_loops : Flag<["-"], "funroll-loops">, Group<f_Group>,
HelpText<"Turn on loop unroller">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
def fno_unroll_loops : Flag<["-"], "fno-unroll-loops">, Group<f_Group>,
diff --git a/clang/lib/CodeGen/BackendUtil.cpp
b/clang/lib/CodeGen/BackendUtil.cpp
index 3f095c03397fd..b16e619727617 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/clang/lib/CodeGen/BackendUtil.cpp
@@ -896,6 +896,7 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
PipelineTuningOptions PTO;
PTO.LoopUnrolling = CodeGenOpts.UnrollLoops;
PTO.LoopInterchange = CodeGenOpts.InterchangeLoops;
+ PTO.LoopFuse = CodeGenOpts.FuseLoops;
// For historical reasons, loop interleaving is set to mirror setting for
loop
// unrolling.
PTO.LoopInterleaving = CodeGenOpts.UnrollLoops;
@@ -1331,6 +1332,7 @@ runThinLTOBackend(CompilerInstance &CI,
ModuleSummaryIndex *CombinedIndex,
Conf.SampleProfile = std::move(SampleProfile);
Conf.PTO.LoopUnrolling = CGOpts.UnrollLoops;
Conf.PTO.LoopInterchange = CGOpts.InterchangeLoops;
+ Conf.PTO.LoopFuse = CGOpts.FuseLoops;
// For historical reasons, loop interleaving is set to mirror setting for
loop
// unrolling.
Conf.PTO.LoopInterleaving = CGOpts.UnrollLoops;
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index 946b1e39af3b9..dcfba95558fe6 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -6854,6 +6854,7 @@ void Clang::ConstructJob(Compilation &C, const JobAction
&JA,
options::OPT_fno_unroll_loops);
Args.AddLastArg(CmdArgs, options::OPT_floop_interchange,
options::OPT_fno_loop_interchange);
+ Args.AddLastArg(CmdArgs, options::OPT_floop_fuse,
options::OPT_fno_loop_fuse);
Args.AddLastArg(CmdArgs, options::OPT_fstrict_flex_arrays_EQ);
diff --git a/clang/lib/Driver/ToolChains/CommonArgs.cpp
b/clang/lib/Driver/ToolChains/CommonArgs.cpp
index b50549219e4e8..a3c75d1009daa 100644
--- a/clang/lib/Driver/ToolChains/CommonArgs.cpp
+++ b/clang/lib/Driver/ToolChains/CommonArgs.cpp
@@ -3334,8 +3334,14 @@ void tools::handleVectorizeSLPArgs(const ArgList &Args,
void tools::handleInterchangeLoopsArgs(const ArgList &Args,
ArgStringList &CmdArgs) {
- if (Args.hasFlag(options::OPT_floop_interchange,
- options::OPT_fno_loop_interchange, false))
+ // FIXME: Instead of relying on shouldEnableVectorizerAtOLevel, we may want
to
+ // implement a separate function to infer loop interchange from opt level.
+ // For now, enable loop-interchange at the same opt levels as loop-vectorize.
+ bool EnableInterchange = shouldEnableVectorizerAtOLevel(Args, false);
+ OptSpecifier InterchangeAliasOption =
+ EnableInterchange ? options::OPT_O_Group :
options::OPT_floop_interchange;
+ if (Args.hasFlag(options::OPT_floop_interchange, InterchangeAliasOption,
+ options::OPT_fno_loop_interchange, EnableInterchange))
CmdArgs.push_back("-floop-interchange");
}
diff --git a/clang/lib/Driver/ToolChains/Flang.cpp
b/clang/lib/Driver/ToolChains/Flang.cpp
index 1535f4cebf436..77bc221d7b709 100644
--- a/clang/lib/Driver/ToolChains/Flang.cpp
+++ b/clang/lib/Driver/ToolChains/Flang.cpp
@@ -151,6 +151,8 @@ void Flang::addCodegenOptions(const ArgList &Args,
!stackArrays->getOption().matches(options::OPT_fno_stack_arrays))
CmdArgs.push_back("-fstack-arrays");
+ Args.AddLastArg(CmdArgs, options::OPT_floop_fuse,
options::OPT_fno_loop_fuse);
+
handleInterchangeLoopsArgs(Args, CmdArgs);
handleVectorizeLoopsArgs(Args, CmdArgs);
handleVectorizeSLPArgs(Args, CmdArgs);
diff --git a/clang/lib/Frontend/CompilerInvocation.cpp
b/clang/lib/Frontend/CompilerInvocation.cpp
index 761310813f787..d6dcc9c24e7ad 100644
--- a/clang/lib/Frontend/CompilerInvocation.cpp
+++ b/clang/lib/Frontend/CompilerInvocation.cpp
@@ -1680,6 +1680,11 @@ void CompilerInvocationBase::GenerateCodeGenArgs(const
CodeGenOptions &Opts,
else
GenerateArg(Consumer, OPT_fno_loop_interchange);
+ if (Opts.FuseLoops)
+ GenerateArg(Consumer, OPT_floop_fuse);
+ else
+ GenerateArg(Consumer, OPT_fno_loop_fuse);
+
if (!Opts.BinutilsVersion.empty())
GenerateArg(Consumer, OPT_fbinutils_version_EQ, Opts.BinutilsVersion);
@@ -2001,6 +2006,7 @@ bool CompilerInvocation::ParseCodeGenArgs(CodeGenOptions
&Opts, ArgList &Args,
(Opts.OptimizationLevel > 1));
Opts.InterchangeLoops =
Args.hasFlag(OPT_floop_interchange, OPT_fno_loop_interchange, false);
+ Opts.FuseLoops = Args.hasFlag(OPT_floop_fuse, OPT_fno_loop_fuse, false);
Opts.BinutilsVersion =
std::string(Args.getLastArgValue(OPT_fbinutils_version_EQ));
diff --git a/clang/test/Driver/clang_f_opts.c b/clang/test/Driver/clang_f_opts.c
index ee7ded265769b..204d5e338190f 100644
--- a/clang/test/Driver/clang_f_opts.c
+++ b/clang/test/Driver/clang_f_opts.c
@@ -52,6 +52,13 @@
// CHECK-INTERCHANGE-LOOPS: "-floop-interchange"
// CHECK-NO-INTERCHANGE-LOOPS: "-fno-loop-interchange"
+// RUN: %clang -### -S -floop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-FUSE-LOOPS %s
+// RUN: %clang -### -S -fno-loop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
+// RUN: %clang -### -S -fno-loop-fuse -floop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-FUSE-LOOPS %s
+// RUN: %clang -### -S -floop-fuse -fno-loop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
+// CHECK-FUSE-LOOPS: "-floop-fuse"
+// CHECK-NO-FUSE-LOOPS: "-fno-loop-fuse"
+
// RUN: %clang -### -S -fprofile-sample-accurate %s 2>&1 | FileCheck
-check-prefix=CHECK-PROFILE-SAMPLE-ACCURATE %s
// CHECK-PROFILE-SAMPLE-ACCURATE: "-fprofile-sample-accurate"
diff --git a/flang/docs/ReleaseNotes.md b/flang/docs/ReleaseNotes.md
index c9623ea08c4e6..516424f689762 100644
--- a/flang/docs/ReleaseNotes.md
+++ b/flang/docs/ReleaseNotes.md
@@ -35,6 +35,10 @@ page](https://llvm.org/releases/).
## New Compiler Flags
+* -floop-interchange is now recognized by flang.
+* -floop-interchange is enabled by default at -O2 and above.
+* -floop-fuse is now recognized by flang.
+
## Windows Support
## Fortran Language Changes in Flang
diff --git a/flang/include/flang/Frontend/CodeGenOptions.def
b/flang/include/flang/Frontend/CodeGenOptions.def
index cdeea93c9aecb..49e4d292a13e1 100644
--- a/flang/include/flang/Frontend/CodeGenOptions.def
+++ b/flang/include/flang/Frontend/CodeGenOptions.def
@@ -43,6 +43,7 @@ CODEGENOPT(StackArrays, 1, 0) ///< -fstack-arrays (enable the
stack-arrays pass)
CODEGENOPT(VectorizeLoop, 1, 0) ///< Enable loop vectorization.
CODEGENOPT(VectorizeSLP, 1, 0) ///< Enable SLP vectorization.
CODEGENOPT(InterchangeLoops, 1, 0) ///< Enable loop interchange.
+CODEGENOPT(FuseLoops, 1, 0) ///< Enable loop fuse.
CODEGENOPT(LoopVersioning, 1, 0) ///< Enable loop versioning.
CODEGENOPT(UnrollLoops, 1, 0) ///< Enable loop unrolling
CODEGENOPT(AliasAnalysis, 1, 0) ///< Enable alias analysis pass
diff --git a/flang/lib/Frontend/CompilerInvocation.cpp
b/flang/lib/Frontend/CompilerInvocation.cpp
index 6295a58b1bdad..d46b4c83e8bf4 100644
--- a/flang/lib/Frontend/CompilerInvocation.cpp
+++ b/flang/lib/Frontend/CompilerInvocation.cpp
@@ -276,6 +276,9 @@ static void
parseCodeGenArgs(Fortran::frontend::CodeGenOptions &opts,
if (args.getLastArg(clang::driver::options::OPT_floop_interchange))
opts.InterchangeLoops = 1;
+ if (args.getLastArg(clang::driver::options::OPT_floop_fuse))
+ opts.FuseLoops = 1;
+
if (args.getLastArg(clang::driver::options::OPT_vectorize_loops))
opts.VectorizeLoop = 1;
diff --git a/flang/lib/Frontend/FrontendActions.cpp
b/flang/lib/Frontend/FrontendActions.cpp
index 3bef6b1c31825..0c031e59e44f6 100644
--- a/flang/lib/Frontend/FrontendActions.cpp
+++ b/flang/lib/Frontend/FrontendActions.cpp
@@ -958,6 +958,7 @@ void
CodeGenAction::runOptimizationPipeline(llvm::raw_pwrite_stream &os) {
si.getTimePasses().setOutStream(ci.getTimingStreamLLVM());
pto.LoopUnrolling = opts.UnrollLoops;
pto.LoopInterchange = opts.InterchangeLoops;
+ pto.LoopFuse = opts.FuseLoops;
pto.LoopInterleaving = opts.UnrollLoops;
pto.LoopVectorization = opts.VectorizeLoop;
pto.SLPVectorization = opts.VectorizeSLP;
diff --git a/flang/test/Driver/loop-fuse.f90 b/flang/test/Driver/loop-fuse.f90
new file mode 100644
index 0000000000000..240d00fdb62d7
--- /dev/null
+++ b/flang/test/Driver/loop-fuse.f90
@@ -0,0 +1,17 @@
+! RUN: %flang -### -S -floop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-LOOP-FUSE %s
+! RUN: %flang -### -S -fno-loop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-LOOP-FUSE %s
+! RUN: %flang -### -S -O0 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
+! RUN: %flang -### -S -O1 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
+! RUN: %flang -### -S -O2 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
+! RUN: %flang -### -S -O3 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
+! RUN: %flang -### -S -Os %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
+! RUN: %flang -### -S -Oz %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
+! CHECK-LOOP-FUSE: "-floop-fuse"
+! CHECK-NO-LOOP-FUSE-NOT: "-floop-fuse"
+! RUN: %flang_fc1 -emit-llvm -O2 -floop-fuse -mllvm -print-pipeline-passes -o
/dev/null %s 2>&1 | FileCheck -check-prefix=CHECK-LOOP-FUSE-PASS %s
+! RUN: %flang_fc1 -emit-llvm -O2 -fno-loop-fuse -mllvm -print-pipeline-passes
-o /dev/null %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE-PASS %s
+! CHECK-LOOP-FUSE-PASS: loop-fusion
+! CHECK-NO-LOOP-FUSE-PASS-NOT: loop-fusion
+
+program test
+end program
diff --git a/llvm/include/llvm/Passes/PassBuilder.h
b/llvm/include/llvm/Passes/PassBuilder.h
index 9cdb7ca7dbc9b..5870f9e6baeac 100644
--- a/llvm/include/llvm/Passes/PassBuilder.h
+++ b/llvm/include/llvm/Passes/PassBuilder.h
@@ -65,6 +65,9 @@ class PipelineTuningOptions {
/// false.
bool LoopInterchange;
+ /// Tuning option to enable/disable loop fuse. Its default value is false.
+ bool LoopFuse;
+
/// Tuning option to forget all SCEV loops in LoopUnroll. Its default value
/// is that of the flag: `-forget-scev-loop-unroll`.
bool ForgetAllSCEVInLoopUnroll;
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 98821bb1408a7..fe077198dfb06 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -104,6 +104,7 @@
#include "llvm/Transforms/Scalar/LoopDeletion.h"
#include "llvm/Transforms/Scalar/LoopDistribute.h"
#include "llvm/Transforms/Scalar/LoopFlatten.h"
+#include "llvm/Transforms/Scalar/LoopFuse.h"
#include "llvm/Transforms/Scalar/LoopIdiomRecognize.h"
#include "llvm/Transforms/Scalar/LoopInstSimplify.h"
#include "llvm/Transforms/Scalar/LoopInterchange.h"
@@ -204,6 +205,10 @@ static cl::opt<bool>
EnableLoopInterchange("enable-loopinterchange", cl::init(false),
cl::Hidden,
cl::desc("Enable the LoopInterchange Pass"));
+static cl::opt<bool> EnableLoopFuse("enable-loopfuse", cl::init(false),
+ cl::Hidden,
+ cl::desc("Enable the LoopFuse Pass"));
+
static cl::opt<bool> EnableUnrollAndJam("enable-unroll-and-jam",
cl::init(false), cl::Hidden,
cl::desc("Enable Unroll And Jam
Pass"));
@@ -313,6 +318,7 @@ PipelineTuningOptions::PipelineTuningOptions() {
SLPVectorization = false;
LoopUnrolling = true;
LoopInterchange = EnableLoopInterchange;
+ LoopFuse = EnableLoopFuse;
ForgetAllSCEVInLoopUnroll = ForgetSCEVInLoopUnroll;
LicmMssaOptCap = SetLicmMssaOptCap;
LicmMssaNoAccForPromotionCap = SetLicmMssaNoAccForPromotionCap;
@@ -514,6 +520,9 @@
PassBuilder::buildO1FunctionSimplificationPipeline(OptimizationLevel Level,
invokeLoopOptimizerEndEPCallbacks(LPM2, Level);
+ if (PTO.LoopFuse)
+ FPM.addPass(LoopFusePass());
+
FPM.addPass(createFunctionToLoopPassAdaptor(std::move(LPM1),
/*UseMemorySSA=*/true,
/*UseBlockFrequencyInfo=*/true));
@@ -703,6 +712,9 @@
PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level,
invokeLoopOptimizerEndEPCallbacks(LPM2, Level);
+ if (PTO.LoopFuse)
+ FPM.addPass(LoopFusePass());
+
FPM.addPass(createFunctionToLoopPassAdaptor(std::move(LPM1),
/*UseMemorySSA=*/true,
/*UseBlockFrequencyInfo=*/true));
@@ -2115,7 +2127,6 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel
Level,
LPM.addPass(LoopFlattenPass());
LPM.addPass(IndVarSimplifyPass());
LPM.addPass(LoopDeletionPass());
- // FIXME: Add loop interchange.
// Unroll small loops and perform peeling.
LPM.addPass(LoopFullUnrollPass(Level.getSpeedupLevel(),
>From 7b4fdc4e4d61e4ad6b9d76c560c1aaa07cb3dcb9 Mon Sep 17 00:00:00 2001
From: Sebastian Pop <s...@nvidia.com>
Date: Thu, 12 Jun 2025 09:36:51 +0000
Subject: [PATCH 02/10] rename -floop-fuse / -fexperimental-fuse-loops
---
clang/include/clang/Basic/CodeGenOptions.def | 2 +-
clang/include/clang/Driver/Options.td | 8 ++++----
clang/lib/CodeGen/BackendUtil.cpp | 4 ++--
clang/lib/Driver/ToolChains/Clang.cpp | 3 ++-
clang/lib/Driver/ToolChains/CommonArgs.cpp | 2 +-
clang/lib/Driver/ToolChains/Flang.cpp | 3 ++-
clang/lib/Frontend/CompilerInvocation.cpp | 7 ++++---
clang/test/Driver/clang_f_opts.c | 12 +++++------
flang/docs/ReleaseNotes.md | 4 +---
.../include/flang/Frontend/CodeGenOptions.def | 2 +-
flang/lib/Frontend/CompilerInvocation.cpp | 2 +-
flang/lib/Frontend/FrontendActions.cpp | 2 +-
flang/test/Driver/loop-fuse.f90 | 12 +++++------
llvm/include/llvm/Passes/PassBuilder.h | 4 ++--
llvm/lib/Passes/PassBuilderPipelines.cpp | 20 +++++++++----------
15 files changed, 43 insertions(+), 44 deletions(-)
diff --git a/clang/include/clang/Basic/CodeGenOptions.def
b/clang/include/clang/Basic/CodeGenOptions.def
index 4c4f0970dfff0..872f73ebf3810 100644
--- a/clang/include/clang/Basic/CodeGenOptions.def
+++ b/clang/include/clang/Basic/CodeGenOptions.def
@@ -322,7 +322,7 @@ CODEGENOPT(TimeTrace , 1, 0, Benign) ///< Set when
-ftime-trace is enabl
VALUE_CODEGENOPT(TimeTraceGranularity, 32, 500, Benign) ///< Minimum time
granularity (in microseconds),
///< traced by time
profiler
CODEGENOPT(InterchangeLoops , 1, 0, Benign) ///< Run loop-interchange.
-CODEGENOPT(FuseLoops , 1, 0, Benign) ///< Run loop-fuse.
+CODEGENOPT(FuseLoops , 1, 0, Benign) ///< Run loop-fusion.
CODEGENOPT(UnrollLoops , 1, 0, Benign) ///< Control whether loops are
unrolled.
CODEGENOPT(RerollLoops , 1, 0, Benign) ///< Control whether loops are
rerolled.
CODEGENOPT(NoUseJumpTables , 1, 0, Benign) ///< Set when -fno-jump-tables is
enabled.
diff --git a/clang/include/clang/Driver/Options.td
b/clang/include/clang/Driver/Options.td
index 13a8502c6c7b5..3ac5bc1e2a477 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -4304,10 +4304,10 @@ def floop_interchange : Flag<["-"],
"floop-interchange">, Group<f_Group>,
HelpText<"Enable the loop interchange pass">, Visibility<[ClangOption,
CC1Option, FlangOption, FC1Option]>;
def fno_loop_interchange: Flag<["-"], "fno-loop-interchange">, Group<f_Group>,
HelpText<"Disable the loop interchange pass">, Visibility<[ClangOption,
CC1Option, FlangOption, FC1Option]>;
-def floop_fuse : Flag<["-"], "floop-fuse">, Group<f_Group>,
- HelpText<"Enable the loop fuse pass">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
-def fno_loop_fuse: Flag<["-"], "fno-loop-fuse">, Group<f_Group>,
- HelpText<"Disable the loop fuse pass">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
+def fexperimental_loop_fusion : Flag<["-"], "fexperimental-loop-fusion">,
Group<f_Group>,
+ HelpText<"Enable the experimental loop fusion pass">,
Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>;
+def fno_experimental_loop_fusion: Flag<["-"], "fno-experimental-loop-fusion">,
Group<f_Group>,
+ HelpText<"Disable the experimental loop fusion pass">,
Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>;
def funroll_loops : Flag<["-"], "funroll-loops">, Group<f_Group>,
HelpText<"Turn on loop unroller">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
def fno_unroll_loops : Flag<["-"], "fno-unroll-loops">, Group<f_Group>,
diff --git a/clang/lib/CodeGen/BackendUtil.cpp
b/clang/lib/CodeGen/BackendUtil.cpp
index b16e619727617..8c99af2bdff83 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/clang/lib/CodeGen/BackendUtil.cpp
@@ -896,7 +896,7 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
PipelineTuningOptions PTO;
PTO.LoopUnrolling = CodeGenOpts.UnrollLoops;
PTO.LoopInterchange = CodeGenOpts.InterchangeLoops;
- PTO.LoopFuse = CodeGenOpts.FuseLoops;
+ PTO.LoopFusion = CodeGenOpts.FuseLoops;
// For historical reasons, loop interleaving is set to mirror setting for
loop
// unrolling.
PTO.LoopInterleaving = CodeGenOpts.UnrollLoops;
@@ -1332,7 +1332,7 @@ runThinLTOBackend(CompilerInstance &CI,
ModuleSummaryIndex *CombinedIndex,
Conf.SampleProfile = std::move(SampleProfile);
Conf.PTO.LoopUnrolling = CGOpts.UnrollLoops;
Conf.PTO.LoopInterchange = CGOpts.InterchangeLoops;
- Conf.PTO.LoopFuse = CGOpts.FuseLoops;
+ Conf.PTO.LoopFusion = CGOpts.FuseLoops;
// For historical reasons, loop interleaving is set to mirror setting for
loop
// unrolling.
Conf.PTO.LoopInterleaving = CGOpts.UnrollLoops;
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index dcfba95558fe6..d3f503c8f6c77 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -6854,7 +6854,8 @@ void Clang::ConstructJob(Compilation &C, const JobAction
&JA,
options::OPT_fno_unroll_loops);
Args.AddLastArg(CmdArgs, options::OPT_floop_interchange,
options::OPT_fno_loop_interchange);
- Args.AddLastArg(CmdArgs, options::OPT_floop_fuse,
options::OPT_fno_loop_fuse);
+ Args.AddLastArg(CmdArgs, options::OPT_fexperimental_loop_fusion,
+ options::OPT_fno_experimental_loop_fusion);
Args.AddLastArg(CmdArgs, options::OPT_fstrict_flex_arrays_EQ);
diff --git a/clang/lib/Driver/ToolChains/CommonArgs.cpp
b/clang/lib/Driver/ToolChains/CommonArgs.cpp
index a3c75d1009daa..bbac9f1af3828 100644
--- a/clang/lib/Driver/ToolChains/CommonArgs.cpp
+++ b/clang/lib/Driver/ToolChains/CommonArgs.cpp
@@ -3334,7 +3334,7 @@ void tools::handleVectorizeSLPArgs(const ArgList &Args,
void tools::handleInterchangeLoopsArgs(const ArgList &Args,
ArgStringList &CmdArgs) {
- // FIXME: Instead of relying on shouldEnableVectorizerAtOLevel, we may want
to
+ // FIXME: instead of relying on shouldEnableVectorizerAtOLevel, we may want
to
// implement a separate function to infer loop interchange from opt level.
// For now, enable loop-interchange at the same opt levels as loop-vectorize.
bool EnableInterchange = shouldEnableVectorizerAtOLevel(Args, false);
diff --git a/clang/lib/Driver/ToolChains/Flang.cpp
b/clang/lib/Driver/ToolChains/Flang.cpp
index 77bc221d7b709..eae914ca68877 100644
--- a/clang/lib/Driver/ToolChains/Flang.cpp
+++ b/clang/lib/Driver/ToolChains/Flang.cpp
@@ -151,7 +151,8 @@ void Flang::addCodegenOptions(const ArgList &Args,
!stackArrays->getOption().matches(options::OPT_fno_stack_arrays))
CmdArgs.push_back("-fstack-arrays");
- Args.AddLastArg(CmdArgs, options::OPT_floop_fuse,
options::OPT_fno_loop_fuse);
+ Args.AddLastArg(CmdArgs, options::OPT_fexperimental_loop_fusion,
+ options::OPT_fno_experimental_loop_fusion);
handleInterchangeLoopsArgs(Args, CmdArgs);
handleVectorizeLoopsArgs(Args, CmdArgs);
diff --git a/clang/lib/Frontend/CompilerInvocation.cpp
b/clang/lib/Frontend/CompilerInvocation.cpp
index d6dcc9c24e7ad..2aec6dc02450c 100644
--- a/clang/lib/Frontend/CompilerInvocation.cpp
+++ b/clang/lib/Frontend/CompilerInvocation.cpp
@@ -1681,9 +1681,9 @@ void CompilerInvocationBase::GenerateCodeGenArgs(const
CodeGenOptions &Opts,
GenerateArg(Consumer, OPT_fno_loop_interchange);
if (Opts.FuseLoops)
- GenerateArg(Consumer, OPT_floop_fuse);
+ GenerateArg(Consumer, OPT_fexperimental_loop_fusion);
else
- GenerateArg(Consumer, OPT_fno_loop_fuse);
+ GenerateArg(Consumer, OPT_fno_experimental_loop_fusion);
if (!Opts.BinutilsVersion.empty())
GenerateArg(Consumer, OPT_fbinutils_version_EQ, Opts.BinutilsVersion);
@@ -2006,7 +2006,8 @@ bool CompilerInvocation::ParseCodeGenArgs(CodeGenOptions
&Opts, ArgList &Args,
(Opts.OptimizationLevel > 1));
Opts.InterchangeLoops =
Args.hasFlag(OPT_floop_interchange, OPT_fno_loop_interchange, false);
- Opts.FuseLoops = Args.hasFlag(OPT_floop_fuse, OPT_fno_loop_fuse, false);
+ Opts.FuseLoops = Args.hasFlag(OPT_fexperimental_loop_fusion,
+ OPT_fno_experimental_loop_fusion, false);
Opts.BinutilsVersion =
std::string(Args.getLastArgValue(OPT_fbinutils_version_EQ));
diff --git a/clang/test/Driver/clang_f_opts.c b/clang/test/Driver/clang_f_opts.c
index 204d5e338190f..ca654cb2dd49b 100644
--- a/clang/test/Driver/clang_f_opts.c
+++ b/clang/test/Driver/clang_f_opts.c
@@ -52,12 +52,12 @@
// CHECK-INTERCHANGE-LOOPS: "-floop-interchange"
// CHECK-NO-INTERCHANGE-LOOPS: "-fno-loop-interchange"
-// RUN: %clang -### -S -floop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-FUSE-LOOPS %s
-// RUN: %clang -### -S -fno-loop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
-// RUN: %clang -### -S -fno-loop-fuse -floop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-FUSE-LOOPS %s
-// RUN: %clang -### -S -floop-fuse -fno-loop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
-// CHECK-FUSE-LOOPS: "-floop-fuse"
-// CHECK-NO-FUSE-LOOPS: "-fno-loop-fuse"
+// RUN: %clang -### -S -fexperimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-FUSE-LOOPS %s
+// RUN: %clang -### -S -fno-experimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
+// RUN: %clang -### -S -fno-experimental-loop-fusion
-fexperimental-loop-fusion %s 2>&1 | FileCheck -check-prefix=CHECK-FUSE-LOOPS %s
+// RUN: %clang -### -S -fexperimental-loop-fusion
-fno-experimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
+// CHECK-FUSE-LOOPS: "-fexperimental-loop-fusion"
+// CHECK-NO-FUSE-LOOPS: "-fno-experimental-loop-fusion"
// RUN: %clang -### -S -fprofile-sample-accurate %s 2>&1 | FileCheck
-check-prefix=CHECK-PROFILE-SAMPLE-ACCURATE %s
// CHECK-PROFILE-SAMPLE-ACCURATE: "-fprofile-sample-accurate"
diff --git a/flang/docs/ReleaseNotes.md b/flang/docs/ReleaseNotes.md
index 516424f689762..6a285f829053b 100644
--- a/flang/docs/ReleaseNotes.md
+++ b/flang/docs/ReleaseNotes.md
@@ -35,9 +35,7 @@ page](https://llvm.org/releases/).
## New Compiler Flags
-* -floop-interchange is now recognized by flang.
-* -floop-interchange is enabled by default at -O2 and above.
-* -floop-fuse is now recognized by flang.
+* -fexperimental-loop-fusion is now recognized by flang.
## Windows Support
diff --git a/flang/include/flang/Frontend/CodeGenOptions.def
b/flang/include/flang/Frontend/CodeGenOptions.def
index 49e4d292a13e1..f273dad9606a6 100644
--- a/flang/include/flang/Frontend/CodeGenOptions.def
+++ b/flang/include/flang/Frontend/CodeGenOptions.def
@@ -43,7 +43,7 @@ CODEGENOPT(StackArrays, 1, 0) ///< -fstack-arrays (enable the
stack-arrays pass)
CODEGENOPT(VectorizeLoop, 1, 0) ///< Enable loop vectorization.
CODEGENOPT(VectorizeSLP, 1, 0) ///< Enable SLP vectorization.
CODEGENOPT(InterchangeLoops, 1, 0) ///< Enable loop interchange.
-CODEGENOPT(FuseLoops, 1, 0) ///< Enable loop fuse.
+CODEGENOPT(FuseLoops, 1, 0) ///< Enable loop fusion.
CODEGENOPT(LoopVersioning, 1, 0) ///< Enable loop versioning.
CODEGENOPT(UnrollLoops, 1, 0) ///< Enable loop unrolling
CODEGENOPT(AliasAnalysis, 1, 0) ///< Enable alias analysis pass
diff --git a/flang/lib/Frontend/CompilerInvocation.cpp
b/flang/lib/Frontend/CompilerInvocation.cpp
index d46b4c83e8bf4..4f42fbd66eac0 100644
--- a/flang/lib/Frontend/CompilerInvocation.cpp
+++ b/flang/lib/Frontend/CompilerInvocation.cpp
@@ -276,7 +276,7 @@ static void
parseCodeGenArgs(Fortran::frontend::CodeGenOptions &opts,
if (args.getLastArg(clang::driver::options::OPT_floop_interchange))
opts.InterchangeLoops = 1;
- if (args.getLastArg(clang::driver::options::OPT_floop_fuse))
+ if (args.getLastArg(clang::driver::options::OPT_fexperimental_loop_fusion))
opts.FuseLoops = 1;
if (args.getLastArg(clang::driver::options::OPT_vectorize_loops))
diff --git a/flang/lib/Frontend/FrontendActions.cpp
b/flang/lib/Frontend/FrontendActions.cpp
index 0c031e59e44f6..23cc1e63e773d 100644
--- a/flang/lib/Frontend/FrontendActions.cpp
+++ b/flang/lib/Frontend/FrontendActions.cpp
@@ -958,7 +958,7 @@ void
CodeGenAction::runOptimizationPipeline(llvm::raw_pwrite_stream &os) {
si.getTimePasses().setOutStream(ci.getTimingStreamLLVM());
pto.LoopUnrolling = opts.UnrollLoops;
pto.LoopInterchange = opts.InterchangeLoops;
- pto.LoopFuse = opts.FuseLoops;
+ pto.LoopFusion = opts.FuseLoops;
pto.LoopInterleaving = opts.UnrollLoops;
pto.LoopVectorization = opts.VectorizeLoop;
pto.SLPVectorization = opts.VectorizeSLP;
diff --git a/flang/test/Driver/loop-fuse.f90 b/flang/test/Driver/loop-fuse.f90
index 240d00fdb62d7..ddfd9065e0fd4 100644
--- a/flang/test/Driver/loop-fuse.f90
+++ b/flang/test/Driver/loop-fuse.f90
@@ -1,15 +1,15 @@
-! RUN: %flang -### -S -floop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-LOOP-FUSE %s
-! RUN: %flang -### -S -fno-loop-fuse %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-LOOP-FUSE %s
+! RUN: %flang -### -S -fexperimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-LOOP-FUSE %s
+! RUN: %flang -### -S -fno-experimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-LOOP-FUSE %s
! RUN: %flang -### -S -O0 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
! RUN: %flang -### -S -O1 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
! RUN: %flang -### -S -O2 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
! RUN: %flang -### -S -O3 %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
! RUN: %flang -### -S -Os %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
! RUN: %flang -### -S -Oz %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE
%s
-! CHECK-LOOP-FUSE: "-floop-fuse"
-! CHECK-NO-LOOP-FUSE-NOT: "-floop-fuse"
-! RUN: %flang_fc1 -emit-llvm -O2 -floop-fuse -mllvm -print-pipeline-passes -o
/dev/null %s 2>&1 | FileCheck -check-prefix=CHECK-LOOP-FUSE-PASS %s
-! RUN: %flang_fc1 -emit-llvm -O2 -fno-loop-fuse -mllvm -print-pipeline-passes
-o /dev/null %s 2>&1 | FileCheck -check-prefix=CHECK-NO-LOOP-FUSE-PASS %s
+! CHECK-LOOP-FUSE: "-fexperimental-loop-fusion"
+! CHECK-NO-LOOP-FUSE-NOT: "-fexperimental-loop-fusion"
+! RUN: %flang_fc1 -emit-llvm -O2 -fexperimental-loop-fusion -mllvm
-print-pipeline-passes -o /dev/null %s 2>&1 | FileCheck
-check-prefix=CHECK-LOOP-FUSE-PASS %s
+! RUN: %flang_fc1 -emit-llvm -O2 -fno-experimental-loop-fusion -mllvm
-print-pipeline-passes -o /dev/null %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-LOOP-FUSE-PASS %s
! CHECK-LOOP-FUSE-PASS: loop-fusion
! CHECK-NO-LOOP-FUSE-PASS-NOT: loop-fusion
diff --git a/llvm/include/llvm/Passes/PassBuilder.h
b/llvm/include/llvm/Passes/PassBuilder.h
index 5870f9e6baeac..2742ec1b71b7e 100644
--- a/llvm/include/llvm/Passes/PassBuilder.h
+++ b/llvm/include/llvm/Passes/PassBuilder.h
@@ -65,8 +65,8 @@ class PipelineTuningOptions {
/// false.
bool LoopInterchange;
- /// Tuning option to enable/disable loop fuse. Its default value is false.
- bool LoopFuse;
+ /// Tuning option to enable/disable loop fusion. Its default value is false.
+ bool LoopFusion;
/// Tuning option to forget all SCEV loops in LoopUnroll. Its default value
/// is that of the flag: `-forget-scev-loop-unroll`.
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index fe077198dfb06..390febda569c3 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -205,9 +205,9 @@ static cl::opt<bool>
EnableLoopInterchange("enable-loopinterchange", cl::init(false),
cl::Hidden,
cl::desc("Enable the LoopInterchange Pass"));
-static cl::opt<bool> EnableLoopFuse("enable-loopfuse", cl::init(false),
- cl::Hidden,
- cl::desc("Enable the LoopFuse Pass"));
+static cl::opt<bool> EnableLoopFusion("enable-loopfusion", cl::init(false),
+ cl::Hidden,
+ cl::desc("Enable the LoopFusion Pass"));
static cl::opt<bool> EnableUnrollAndJam("enable-unroll-and-jam",
cl::init(false), cl::Hidden,
@@ -318,7 +318,7 @@ PipelineTuningOptions::PipelineTuningOptions() {
SLPVectorization = false;
LoopUnrolling = true;
LoopInterchange = EnableLoopInterchange;
- LoopFuse = EnableLoopFuse;
+ LoopFusion = EnableLoopFusion;
ForgetAllSCEVInLoopUnroll = ForgetSCEVInLoopUnroll;
LicmMssaOptCap = SetLicmMssaOptCap;
LicmMssaNoAccForPromotionCap = SetLicmMssaNoAccForPromotionCap;
@@ -520,9 +520,6 @@
PassBuilder::buildO1FunctionSimplificationPipeline(OptimizationLevel Level,
invokeLoopOptimizerEndEPCallbacks(LPM2, Level);
- if (PTO.LoopFuse)
- FPM.addPass(LoopFusePass());
-
FPM.addPass(createFunctionToLoopPassAdaptor(std::move(LPM1),
/*UseMemorySSA=*/true,
/*UseBlockFrequencyInfo=*/true));
@@ -647,6 +644,9 @@
PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level,
if (EnableConstraintElimination)
FPM.addPass(ConstraintEliminationPass());
+ if (PTO.LoopFusion)
+ FPM.addPass(LoopFusePass());
+
// Add the primary loop simplification pipeline.
// FIXME: Currently this is split into two loop pass pipelines because we run
// some function passes in between them. These can and should be removed
@@ -712,9 +712,6 @@
PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level,
invokeLoopOptimizerEndEPCallbacks(LPM2, Level);
- if (PTO.LoopFuse)
- FPM.addPass(LoopFusePass());
-
FPM.addPass(createFunctionToLoopPassAdaptor(std::move(LPM1),
/*UseMemorySSA=*/true,
/*UseBlockFrequencyInfo=*/true));
@@ -2127,6 +2124,7 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel
Level,
LPM.addPass(LoopFlattenPass());
LPM.addPass(IndVarSimplifyPass());
LPM.addPass(LoopDeletionPass());
+ // FIXME: Add loop interchange.
// Unroll small loops and perform peeling.
LPM.addPass(LoopFullUnrollPass(Level.getSpeedupLevel(),
@@ -2366,4 +2364,4 @@ AAManager PassBuilder::buildDefaultAAPipeline() {
bool PassBuilder::isInstrumentedPGOUse() const {
return (PGOOpt && PGOOpt->Action == PGOOptions::IRUse) ||
!UseCtxProfile.empty();
-}
\ No newline at end of file
+}
>From fe53e398179ec3d4f85e150b1fd829324f50e816 Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Thu, 31 Jul 2025 09:09:30 -0700
Subject: [PATCH 03/10] Address review comments
---
llvm/lib/Passes/PassBuilderPipelines.cpp | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 390febda569c3..9dfe0180de0c4 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -207,7 +207,7 @@ static cl::opt<bool>
static cl::opt<bool> EnableLoopFusion("enable-loopfusion", cl::init(false),
cl::Hidden,
- cl::desc("Enable the LoopFusion Pass"));
+ cl::desc("Enable the LoopFuse Pass"));
static cl::opt<bool> EnableUnrollAndJam("enable-unroll-and-jam",
cl::init(false), cl::Hidden,
@@ -644,6 +644,8 @@
PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level,
if (EnableConstraintElimination)
FPM.addPass(ConstraintEliminationPass());
+ // FIXME: This may not be the right place in the pipeline.
+ // We need to have the data to support the right place.
if (PTO.LoopFusion)
FPM.addPass(LoopFusePass());
>From ab202f026e6505d35b63943b28dbd0b6fb7d08be Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Thu, 7 Aug 2025 06:42:52 -0700
Subject: [PATCH 04/10] change place of pass in pipeline
---
llvm/lib/Passes/PassBuilderPipelines.cpp | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 9dfe0180de0c4..2df23c5dcf299 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -644,11 +644,6 @@
PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level,
if (EnableConstraintElimination)
FPM.addPass(ConstraintEliminationPass());
- // FIXME: This may not be the right place in the pipeline.
- // We need to have the data to support the right place.
- if (PTO.LoopFusion)
- FPM.addPass(LoopFusePass());
-
// Add the primary loop simplification pipeline.
// FIXME: Currently this is split into two loop pass pipelines because we run
// some function passes in between them. These can and should be removed
@@ -1559,6 +1554,11 @@
PassBuilder::buildModuleOptimizationPipeline(OptimizationLevel Level,
if (PTO.LoopInterchange)
LPM.addPass(LoopInterchangePass());
+ // FIXME: This may not be the right place in the pipeline.
+ // We need to have the data to support the right place.
+ if (PTO.LoopFusion)
+ OptimizePM.addPass(LoopFusePass());
+
OptimizePM.addPass(createFunctionToLoopPassAdaptor(
std::move(LPM), /*UseMemorySSA=*/false,
/*UseBlockFrequencyInfo=*/false));
>From 3f88802809e8bcaafcb63b47887899bd3d34d7ab Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Fri, 8 Aug 2025 04:17:44 -0700
Subject: [PATCH 05/10] fixup! change place of pass in pipeline
also add a RUN line for testing position in pass pipeline
---
clang/include/clang/Driver/Options.td | 8 ++++----
clang/lib/Frontend/CompilerInvocation.cpp | 2 --
clang/test/Driver/clang_f_opts.c | 6 ++++++
llvm/lib/Passes/PassBuilderPipelines.cpp | 6 +++---
4 files changed, 13 insertions(+), 9 deletions(-)
diff --git a/clang/include/clang/Driver/Options.td
b/clang/include/clang/Driver/Options.td
index 3ac5bc1e2a477..3b58267833009 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -4304,10 +4304,10 @@ def floop_interchange : Flag<["-"],
"floop-interchange">, Group<f_Group>,
HelpText<"Enable the loop interchange pass">, Visibility<[ClangOption,
CC1Option, FlangOption, FC1Option]>;
def fno_loop_interchange: Flag<["-"], "fno-loop-interchange">, Group<f_Group>,
HelpText<"Disable the loop interchange pass">, Visibility<[ClangOption,
CC1Option, FlangOption, FC1Option]>;
-def fexperimental_loop_fusion : Flag<["-"], "fexperimental-loop-fusion">,
Group<f_Group>,
- HelpText<"Enable the experimental loop fusion pass">,
Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>;
-def fno_experimental_loop_fusion: Flag<["-"], "fno-experimental-loop-fusion">,
Group<f_Group>,
- HelpText<"Disable the experimental loop fusion pass">,
Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>;
+defm experimental_loop_fusion
+ : OptInCC1FFlag<"experimental-loop-fusion", "Enable", "Disable",
+ "Enable the loop fusion pass",
+ [ClangOption, CC1Option, FlangOption, FC1Option]>;
def funroll_loops : Flag<["-"], "funroll-loops">, Group<f_Group>,
HelpText<"Turn on loop unroller">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
def fno_unroll_loops : Flag<["-"], "fno-unroll-loops">, Group<f_Group>,
diff --git a/clang/lib/Frontend/CompilerInvocation.cpp
b/clang/lib/Frontend/CompilerInvocation.cpp
index 2aec6dc02450c..422375240bab6 100644
--- a/clang/lib/Frontend/CompilerInvocation.cpp
+++ b/clang/lib/Frontend/CompilerInvocation.cpp
@@ -1682,8 +1682,6 @@ void CompilerInvocationBase::GenerateCodeGenArgs(const
CodeGenOptions &Opts,
if (Opts.FuseLoops)
GenerateArg(Consumer, OPT_fexperimental_loop_fusion);
- else
- GenerateArg(Consumer, OPT_fno_experimental_loop_fusion);
if (!Opts.BinutilsVersion.empty())
GenerateArg(Consumer, OPT_fbinutils_version_EQ, Opts.BinutilsVersion);
diff --git a/clang/test/Driver/clang_f_opts.c b/clang/test/Driver/clang_f_opts.c
index ca654cb2dd49b..a9e8414db37a4 100644
--- a/clang/test/Driver/clang_f_opts.c
+++ b/clang/test/Driver/clang_f_opts.c
@@ -58,6 +58,12 @@
// RUN: %clang -### -S -fexperimental-loop-fusion
-fno-experimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
// CHECK-FUSE-LOOPS: "-fexperimental-loop-fusion"
// CHECK-NO-FUSE-LOOPS: "-fno-experimental-loop-fusion"
+//
+// RUN: %clang -c -fexperimental-loop-fusion -mllvm -print-pipeline-passes -O3
%s 2>&1 | FileCheck --check-prefixes=LOOP-FUSION-ON %s
+// RUN: %clang -c -mllvm -print-pipeline-passes -O3 %s 2>&1 | FileCheck
--check-prefixes=LOOP-FUSION-OFF %s
+
+// LOOP-FUSION-ON: loop-fusion
+// LOOP-FUSION-OFF-NOT: loop-fusion
// RUN: %clang -### -S -fprofile-sample-accurate %s 2>&1 | FileCheck
-check-prefix=CHECK-PROFILE-SAMPLE-ACCURATE %s
// CHECK-PROFILE-SAMPLE-ACCURATE: "-fprofile-sample-accurate"
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 2df23c5dcf299..f3e3fb2eafccb 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -1554,14 +1554,14 @@
PassBuilder::buildModuleOptimizationPipeline(OptimizationLevel Level,
if (PTO.LoopInterchange)
LPM.addPass(LoopInterchangePass());
+ OptimizePM.addPass(createFunctionToLoopPassAdaptor(
+ std::move(LPM), /*UseMemorySSA=*/false,
/*UseBlockFrequencyInfo=*/false));
+
// FIXME: This may not be the right place in the pipeline.
// We need to have the data to support the right place.
if (PTO.LoopFusion)
OptimizePM.addPass(LoopFusePass());
- OptimizePM.addPass(createFunctionToLoopPassAdaptor(
- std::move(LPM), /*UseMemorySSA=*/false,
/*UseBlockFrequencyInfo=*/false));
-
// Distribute loops to allow partial vectorization. I.e. isolate dependences
// into separate loop that would otherwise inhibit vectorization. This is
// currently only performed for loops marked with the metadata
>From 9f264b4688ed54c9304777c5b3812a8141f58884 Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Thu, 28 Aug 2025 01:43:38 -0700
Subject: [PATCH 06/10] fixup! address review comments about optInFlag
---
clang/include/clang/Driver/Options.td | 2 +-
clang/lib/Driver/ToolChains/Clang.cpp | 2 +-
clang/test/Driver/clang_f_opts.c | 4 ----
3 files changed, 2 insertions(+), 6 deletions(-)
diff --git a/clang/include/clang/Driver/Options.td
b/clang/include/clang/Driver/Options.td
index 3b58267833009..47d328f862e07 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -4307,7 +4307,7 @@ def fno_loop_interchange: Flag<["-"],
"fno-loop-interchange">, Group<f_Group>,
defm experimental_loop_fusion
: OptInCC1FFlag<"experimental-loop-fusion", "Enable", "Disable",
"Enable the loop fusion pass",
- [ClangOption, CC1Option, FlangOption, FC1Option]>;
+ [ClangOption, FlangOption, FC1Option]>;
def funroll_loops : Flag<["-"], "funroll-loops">, Group<f_Group>,
HelpText<"Turn on loop unroller">, Visibility<[ClangOption, CC1Option,
FlangOption, FC1Option]>;
def fno_unroll_loops : Flag<["-"], "fno-unroll-loops">, Group<f_Group>,
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index d3f503c8f6c77..dfcf06727847c 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -6854,7 +6854,7 @@ void Clang::ConstructJob(Compilation &C, const JobAction
&JA,
options::OPT_fno_unroll_loops);
Args.AddLastArg(CmdArgs, options::OPT_floop_interchange,
options::OPT_fno_loop_interchange);
- Args.AddLastArg(CmdArgs, options::OPT_fexperimental_loop_fusion,
+ Args.addOptInFlag(CmdArgs, options::OPT_fexperimental_loop_fusion,
options::OPT_fno_experimental_loop_fusion);
Args.AddLastArg(CmdArgs, options::OPT_fstrict_flex_arrays_EQ);
diff --git a/clang/test/Driver/clang_f_opts.c b/clang/test/Driver/clang_f_opts.c
index a9e8414db37a4..eb3994ddabcd3 100644
--- a/clang/test/Driver/clang_f_opts.c
+++ b/clang/test/Driver/clang_f_opts.c
@@ -53,11 +53,7 @@
// CHECK-NO-INTERCHANGE-LOOPS: "-fno-loop-interchange"
// RUN: %clang -### -S -fexperimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-FUSE-LOOPS %s
-// RUN: %clang -### -S -fno-experimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
-// RUN: %clang -### -S -fno-experimental-loop-fusion
-fexperimental-loop-fusion %s 2>&1 | FileCheck -check-prefix=CHECK-FUSE-LOOPS %s
-// RUN: %clang -### -S -fexperimental-loop-fusion
-fno-experimental-loop-fusion %s 2>&1 | FileCheck
-check-prefix=CHECK-NO-FUSE-LOOPS %s
// CHECK-FUSE-LOOPS: "-fexperimental-loop-fusion"
-// CHECK-NO-FUSE-LOOPS: "-fno-experimental-loop-fusion"
//
// RUN: %clang -c -fexperimental-loop-fusion -mllvm -print-pipeline-passes -O3
%s 2>&1 | FileCheck --check-prefixes=LOOP-FUSION-ON %s
// RUN: %clang -c -mllvm -print-pipeline-passes -O3 %s 2>&1 | FileCheck
--check-prefixes=LOOP-FUSION-OFF %s
>From 76780b04b1c718acf8be5f77b031c85316834818 Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Thu, 28 Aug 2025 10:04:33 -0700
Subject: [PATCH 07/10] fixup! addOptInFlag for Flang.cpp too.
---
clang/lib/Driver/ToolChains/Flang.cpp | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/clang/lib/Driver/ToolChains/Flang.cpp
b/clang/lib/Driver/ToolChains/Flang.cpp
index eae914ca68877..d3f4af164f672 100644
--- a/clang/lib/Driver/ToolChains/Flang.cpp
+++ b/clang/lib/Driver/ToolChains/Flang.cpp
@@ -151,8 +151,8 @@ void Flang::addCodegenOptions(const ArgList &Args,
!stackArrays->getOption().matches(options::OPT_fno_stack_arrays))
CmdArgs.push_back("-fstack-arrays");
- Args.AddLastArg(CmdArgs, options::OPT_fexperimental_loop_fusion,
- options::OPT_fno_experimental_loop_fusion);
+ Args.addOptInFlag(CmdArgs, options::OPT_fexperimental_loop_fusion,
+ options::OPT_fno_experimental_loop_fusion);
handleInterchangeLoopsArgs(Args, CmdArgs);
handleVectorizeLoopsArgs(Args, CmdArgs);
>From 00d5ca55245daf6990905730461f66e34ca7e1ae Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Thu, 28 Aug 2025 22:09:03 -0700
Subject: [PATCH 08/10] fixup! alignments
---
clang/lib/Driver/ToolChains/Clang.cpp | 2 +-
flang/include/flang/Frontend/CodeGenOptions.def | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index dfcf06727847c..63efb0f02baa8 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -6855,7 +6855,7 @@ void Clang::ConstructJob(Compilation &C, const JobAction
&JA,
Args.AddLastArg(CmdArgs, options::OPT_floop_interchange,
options::OPT_fno_loop_interchange);
Args.addOptInFlag(CmdArgs, options::OPT_fexperimental_loop_fusion,
- options::OPT_fno_experimental_loop_fusion);
+ options::OPT_fno_experimental_loop_fusion);
Args.AddLastArg(CmdArgs, options::OPT_fstrict_flex_arrays_EQ);
diff --git a/flang/include/flang/Frontend/CodeGenOptions.def
b/flang/include/flang/Frontend/CodeGenOptions.def
index f273dad9606a6..edab48a70d29d 100644
--- a/flang/include/flang/Frontend/CodeGenOptions.def
+++ b/flang/include/flang/Frontend/CodeGenOptions.def
@@ -43,7 +43,7 @@ CODEGENOPT(StackArrays, 1, 0) ///< -fstack-arrays (enable the
stack-arrays pass)
CODEGENOPT(VectorizeLoop, 1, 0) ///< Enable loop vectorization.
CODEGENOPT(VectorizeSLP, 1, 0) ///< Enable SLP vectorization.
CODEGENOPT(InterchangeLoops, 1, 0) ///< Enable loop interchange.
-CODEGENOPT(FuseLoops, 1, 0) ///< Enable loop fusion.
+CODEGENOPT(FuseLoops, 1, 0) ///< Enable loop fusion.
CODEGENOPT(LoopVersioning, 1, 0) ///< Enable loop versioning.
CODEGENOPT(UnrollLoops, 1, 0) ///< Enable loop unrolling
CODEGENOPT(AliasAnalysis, 1, 0) ///< Enable alias analysis pass
>From 551813b0687c6a39373ff79743dc8a81db163650 Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Fri, 29 Aug 2025 10:20:22 -0700
Subject: [PATCH 09/10] fixup! add check for EnableFusion flag
---
llvm/lib/Passes/PassBuilderPipelines.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index f3e3fb2eafccb..2f69b40e0bc69 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -1559,7 +1559,7 @@
PassBuilder::buildModuleOptimizationPipeline(OptimizationLevel Level,
// FIXME: This may not be the right place in the pipeline.
// We need to have the data to support the right place.
- if (PTO.LoopFusion)
+ if (PTO.LoopFusion || EnableLoopFusion)
OptimizePM.addPass(LoopFusePass());
// Distribute loops to allow partial vectorization. I.e. isolate dependences
>From a007cea14e9affac6f632f811c9eea3a43c9e15e Mon Sep 17 00:00:00 2001
From: Madhur Amilkanthwar <madh...@nvidia.com>
Date: Mon, 1 Sep 2025 01:47:43 -0700
Subject: [PATCH 10/10] fixup! move enable-fusion to NewPMDriver
---
llvm/lib/Passes/PassBuilderPipelines.cpp | 7 +------
llvm/tools/opt/NewPMDriver.cpp | 4 ++++
2 files changed, 5 insertions(+), 6 deletions(-)
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 2f69b40e0bc69..79642e650ac83 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -205,10 +205,6 @@ static cl::opt<bool>
EnableLoopInterchange("enable-loopinterchange", cl::init(false),
cl::Hidden,
cl::desc("Enable the LoopInterchange Pass"));
-static cl::opt<bool> EnableLoopFusion("enable-loopfusion", cl::init(false),
- cl::Hidden,
- cl::desc("Enable the LoopFuse Pass"));
-
static cl::opt<bool> EnableUnrollAndJam("enable-unroll-and-jam",
cl::init(false), cl::Hidden,
cl::desc("Enable Unroll And Jam
Pass"));
@@ -318,7 +314,6 @@ PipelineTuningOptions::PipelineTuningOptions() {
SLPVectorization = false;
LoopUnrolling = true;
LoopInterchange = EnableLoopInterchange;
- LoopFusion = EnableLoopFusion;
ForgetAllSCEVInLoopUnroll = ForgetSCEVInLoopUnroll;
LicmMssaOptCap = SetLicmMssaOptCap;
LicmMssaNoAccForPromotionCap = SetLicmMssaNoAccForPromotionCap;
@@ -1559,7 +1554,7 @@
PassBuilder::buildModuleOptimizationPipeline(OptimizationLevel Level,
// FIXME: This may not be the right place in the pipeline.
// We need to have the data to support the right place.
- if (PTO.LoopFusion || EnableLoopFusion)
+ if (PTO.LoopFusion)
OptimizePM.addPass(LoopFusePass());
// Distribute loops to allow partial vectorization. I.e. isolate dependences
diff --git a/llvm/tools/opt/NewPMDriver.cpp b/llvm/tools/opt/NewPMDriver.cpp
index b9b8929a0f703..0c991b71a6b26 100644
--- a/llvm/tools/opt/NewPMDriver.cpp
+++ b/llvm/tools/opt/NewPMDriver.cpp
@@ -60,6 +60,9 @@ cl::opt<bool> VerifyEachDebugInfoPreserve(
cl::desc("Start each pass with collecting and end it with checking of "
"debug info preservation."));
+static cl::opt<bool> EnableLoopFusion("enable-loopfusion", cl::init(false),
+ cl::Hidden,
+ cl::desc("Enable the LoopFuse Pass"));
cl::opt<std::string>
VerifyDIPreserveExport("verify-di-preserve-export",
cl::desc("Export debug info preservation failures into "
@@ -446,6 +449,7 @@ bool llvm::runPassPipeline(
// option has been enabled.
PTO.LoopUnrolling = !DisableLoopUnrolling;
PTO.UnifiedLTO = UnifiedLTO;
+ PTO.LoopFusion = EnableLoopFusion;
PassBuilder PB(TM, PTO, P, &PIC);
registerEPCallbacks(PB);
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
