This is an archive of the discontinued LLVM Phabricator instance.

Differential D106799

[OpenMP] Always inline the OpenMP outlined function
ClosedPublic

Authored by jhuber6 on Jul 26 2021, 8:36 AM.

Details

Reviewers
jdoerfert
Commits
rGaf000197c421: [OpenMP] Always inline the OpenMP outlined function
Summary

This patch adds the always inline attribute to the outlined functions generated
by OpenMP regions. Because there is only a single instance of this function and
it always has internal linkage it is safe to inline in every instance it is
created. This could potentially lead to performance degredation due to
inflated register counts in the parallel region.

Diff Detail

Unit TestsFailed

TimeTest
3,020 msx64 debian > libarcher.barrier::barrier.c
2,980 msx64 debian > libarcher.critical::critical.c
3,630 msx64 debian > libarcher.parallel::parallel-simple2.c
3,350 msx64 debian > libarcher.races::critical-unrelated.c
3,450 msx64 debian > libarcher.races::lock-nested-unrelated.c
View Full Test Results (22 Failed)

Event Timeline

jhuber6 created this revision.Jul 26 2021, 8:36 AM
Herald added subscribers: guansong, yaxunl. · View Herald TranscriptJul 26 2021, 8:36 AM
jhuber6 requested review of this revision.Jul 26 2021, 8:36 AM
Herald added a project: Restricted Project. · View Herald TranscriptJul 26 2021, 8:36 AM
Herald added subscribers: cfe-commits, sstefan1. · View Herald Transcript
Harbormaster completed remote builds in B116191: Diff 361676.Jul 26 2021, 9:25 AM
jdoerfert accepted this revision.Jul 26 2021, 12:00 PM

One test needs to be fixed. LG

This revision is now accepted and ready to land.Jul 26 2021, 12:00 PM
jhuber6 updated this revision to Diff 361764.Jul 26 2021, 12:37 PM

Fix test

Harbormaster completed remote builds in B116250: Diff 361764.Jul 26 2021, 1:42 PM
This revision was landed with ongoing or failed builds.Jul 26 2021, 2:28 PM
Closed by commit rGaf000197c421: [OpenMP] Always inline the OpenMP outlined function (authored by jhuber6). · Explain Why
This revision was automatically updated to reflect the committed changes.
jhuber6 added a commit: rGaf000197c421: [OpenMP] Always inline the OpenMP outlined function.
mikerice added a subscriber: mikerice.Aug 4 2021, 4:35 PM

Looks like now omp.outlined functions are marked with both 'noinline' and 'alwaysinline'. See https://bugs.llvm.org/show_bug.cgi?id=51349

jhuber6 mentioned this in D107649: [OpenMP]Fix PR51349: Remove AlwaysInline for if regions..Aug 6 2021, 8:01 AM
jhuber6 mentioned this in rG41a6b50c2596: [OpenMP]Fix PR51349: Remove AlwaysInline for if regions..Aug 6 2021, 2:53 PM
peixin mentioned this in D107430: [OMPIRBuilder] Add ordered directive to OMPIRBuilder.Aug 17 2021, 6:27 AM
igor.kirillov mentioned this in D112645: [OpenMP] Fix: opposite attributes could be set by -fno-inline.Oct 27 2021, 11:46 AM
igor.kirillov mentioned this in rG4860f6cb25bd: [OpenMP] Fix: opposite attributes could be set by -fno-inline.Nov 10 2021, 8:56 AM

Revision Contents

PathSize
clang/
lib/
CodeGen/
6 lines
test/
OpenMP/
147 lines

Diff 361764

clang/lib/CodeGen/CGStmtOpenMP.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 511 Lines • ▼ Show 20 Linesstatic llvm::Function *emitOutlinedFunctionPrologue(
auto *F = auto *F =
llvm::Function::Create(FuncLLVMTy, llvm::GlobalValue::InternalLinkage, llvm::Function::Create(FuncLLVMTy, llvm::GlobalValue::InternalLinkage,
FO.FunctionName, &CGM.getModule()); FO.FunctionName, &CGM.getModule());
CGM.SetInternalFunctionAttributes(CD, F, FuncInfo); CGM.SetInternalFunctionAttributes(CD, F, FuncInfo);
if (CD->isNothrow()) if (CD->isNothrow())
F->setDoesNotThrow(); F->setDoesNotThrow();
F->setDoesNotRecurse(); F->setDoesNotRecurse();
// Always inline the outlined function if optimizations are enabled.
if (CGM.getCodeGenOpts().OptimizationLevel != 0)
F->addFnAttr(llvm::Attribute::AlwaysInline);
// Generate the function. // Generate the function.
CGF.StartFunction(CD, Ctx.VoidTy, F, FuncInfo, TargetArgs, CGF.StartFunction(CD, Ctx.VoidTy, F, FuncInfo, TargetArgs,
FO.UIntPtrCastRequired ? FO.Loc : FO.S->getBeginLoc(), FO.UIntPtrCastRequired ? FO.Loc : FO.S->getBeginLoc(),
FO.UIntPtrCastRequired ? FO.Loc FO.UIntPtrCastRequired ? FO.Loc
: CD->getBody()->getBeginLoc()); : CD->getBody()->getBeginLoc());
unsigned Cnt = CD->getContextParamPosition(); unsigned Cnt = CD->getContextParamPosition();
I = FO.S->captures().begin(); I = FO.S->captures().begin();
for (const FieldDecl *FD : RD->fields()) { for (const FieldDecl *FD : RD->fields()) {
▲ Show 20 Lines • Show All 4,770 Lines • ▼ Show 20 Lines
static llvm::Function *emitOutlinedOrderedFunction(CodeGenModule &CGM, static llvm::Function *emitOutlinedOrderedFunction(CodeGenModule &CGM,
const CapturedStmt *S, const CapturedStmt *S,
SourceLocation Loc) { SourceLocation Loc) {
CodeGenFunction CGF(CGM, /*suppressNewContext=*/true); CodeGenFunction CGF(CGM, /*suppressNewContext=*/true);
CodeGenFunction::CGCapturedStmtInfo CapStmtInfo; CodeGenFunction::CGCapturedStmtInfo CapStmtInfo;
CGF.CapturedStmtInfo = &CapStmtInfo; CGF.CapturedStmtInfo = &CapStmtInfo;
llvm::Function *Fn = CGF.GenerateOpenMPCapturedStmtFunction(*S, Loc); llvm::Function *Fn = CGF.GenerateOpenMPCapturedStmtFunction(*S, Loc);
Fn->setDoesNotRecurse(); Fn->setDoesNotRecurse();
if (CGM.getCodeGenOpts().OptimizationLevel != 0)
Fn->addFnAttr(llvm::Attribute::AlwaysInline);
return Fn; return Fn;
} }
void CodeGenFunction::EmitOMPOrderedDirective(const OMPOrderedDirective &S) { void CodeGenFunction::EmitOMPOrderedDirective(const OMPOrderedDirective &S) {
if (S.hasClausesOfKind<OMPDependClause>()) { if (S.hasClausesOfKind<OMPDependClause>()) {
assert(!S.hasAssociatedStmt() && assert(!S.hasAssociatedStmt() &&
"No associated statement must be in ordered depend construct."); "No associated statement must be in ordered depend construct.");
for (const auto *DC : S.getClausesOfKind<OMPDependClause>()) for (const auto *DC : S.getClausesOfKind<OMPDependClause>())
▲ Show 20 Lines • Show All 1,876 Lines • Show Last 20 Lines

clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
// RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc // RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple nvptx64-unknown-unknown -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK1 // RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple nvptx64-unknown-unknown -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple i386-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm-bc %s -o %t-x86-host.bc // RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple i386-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm-bc %s -o %t-x86-host.bc
// RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple nvptx64-unknown-unknown -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK2 // RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -triple nvptx64-unknown-unknown -aux-triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK2
// RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -fexceptions -fcxx-exceptions -aux-triple powerpc64le-unknown-unknown -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK3 // RUN: %clang_cc1 -x c++ -O1 -disable-llvm-optzns -verify -fopenmp -internal-isystem %S/../Headers/Inputs/include -internal-isystem %S/../../lib/Headers/openmp_wrappers -include __clang_openmp_device_functions.h -fexceptions -fcxx-exceptions -aux-triple powerpc64le-unknown-unknown -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK3
// expected-no-diagnostics // expected-no-diagnostics
#ifndef HEADER #ifndef HEADER
#define HEADER #define HEADER
#include <complex> #include <complex>
// Verify we do not add tbaa metadata to type punned memory operations: // Verify we do not add tbaa metadata to type punned memory operations:
template <typename T> template <typename T>
void complex_reduction() { void complex_reduction() {
#pragma omp target teams distribute #pragma omp target teams distribute
for (int ib = 0; ib < 100; ib++) { for (int ib = 0; ib < 100; ib++) {
std::complex<T> partial_sum; std::complex<T> partial_sum;
const int istart = ib * 4; const int istart = ib * 4;
const int iend = (ib + 1) * 4; const int iend = (ib + 1) * 4;
#pragma omp parallel for reduction(+ \ #pragma omp parallel for reduction(+ \
: partial_sum) : partial_sum)
for (int i = istart; i < iend; i++) for (int i = istart; i < iend; i++)
partial_sum += std::complex<T>(i, i); partial_sum += std::complex<T>(i, i);
} }
} }
void test() { void test() {
complex_reduction<float>(); complex_reduction<float>();
complex_reduction<double>(); complex_reduction<double>();
} }
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l19 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] { // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i1 false, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i1 false, i1 true, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[TMP14:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP14:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP14]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP14]]) #[[ATTR1]]
// CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14:![0-9]+]] // CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14:![0-9]+]]
// CHECK1-NEXT: [[TMP15:%.*]] = bitcast float* [[REF_TMP2]] to i8* // CHECK1-NEXT: [[TMP15:%.*]] = bitcast float* [[REF_TMP2]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP15]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP15]]) #[[ATTR1]]
// CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP2]], align 4, !tbaa [[TBAA14]] // CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP2]], align 4, !tbaa [[TBAA14]]
// CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM_ON_STACK]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP2]]) #[[ATTR8:[0-9]+]] // CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM_ON_STACK]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP2]]) #[[ATTR9:[0-9]+]]
// CHECK1-NEXT: [[TMP16:%.*]] = bitcast float* [[REF_TMP2]] to i8* // CHECK1-NEXT: [[TMP16:%.*]] = bitcast float* [[REF_TMP2]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP16]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP16]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP17:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP17]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP17]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4 // CHECK1-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4
// CHECK1-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[__RE_ADDR:%.*]] = alloca float*, align 8 // CHECK1-NEXT: [[__RE_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[__IM_ADDR:%.*]] = alloca float*, align 8 // CHECK1-NEXT: [[__IM_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store float* [[__RE]], float** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store float* [[__RE]], float** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store float* [[__IM]], float** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store float* [[__IM]], float** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load float*, float** [[__RE_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load float*, float** [[__RE_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load float*, float** [[__IM_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load float*, float** [[__IM_ADDR]], align 8
// CHECK1-NEXT: call void @_ZNSt7complexIfEC2ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS1]], float* nonnull align 4 dereferenceable(4) [[TMP0]], float* nonnull align 4 dereferenceable(4) [[TMP1]]) #[[ATTR8]] // CHECK1-NEXT: call void @_ZNSt7complexIfEC2ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS1]], float* nonnull align 4 dereferenceable(4) [[TMP0]], float* nonnull align 4 dereferenceable(4) [[TMP1]]) #[[ATTR9]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK1-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP21]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP21]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP22:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP22:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP22]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP22]]) #[[ATTR1]]
// CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14]] // CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14]]
// CHECK1-NEXT: [[TMP23:%.*]] = bitcast float* [[REF_TMP6]] to i8* // CHECK1-NEXT: [[TMP23:%.*]] = bitcast float* [[REF_TMP6]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP23]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP23]]) #[[ATTR1]]
// CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP6]], align 4, !tbaa [[TBAA14]] // CHECK1-NEXT: store float 0.000000e+00, float* [[REF_TMP6]], align 4, !tbaa [[TBAA14]]
// CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP6]]) #[[ATTR8]] // CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP6]]) #[[ATTR9]]
// CHECK1-NEXT: [[TMP24:%.*]] = bitcast float* [[REF_TMP6]] to i8* // CHECK1-NEXT: [[TMP24:%.*]] = bitcast float* [[REF_TMP6]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP24]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP24]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP25:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP25]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP25]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8* // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to float // CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to float
// CHECK1-NEXT: store float [[CONV]], float* [[REF_TMP15]], align 4, !tbaa [[TBAA14]] // CHECK1-NEXT: store float [[CONV]], float* [[REF_TMP15]], align 4, !tbaa [[TBAA14]]
// CHECK1-NEXT: [[TMP43:%.*]] = bitcast float* [[REF_TMP16]] to i8* // CHECK1-NEXT: [[TMP43:%.*]] = bitcast float* [[REF_TMP16]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP43]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP43]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to float // CHECK1-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to float
// CHECK1-NEXT: store float [[CONV17]], float* [[REF_TMP16]], align 4, !tbaa [[TBAA14]] // CHECK1-NEXT: store float [[CONV17]], float* [[REF_TMP16]], align 4, !tbaa [[TBAA14]]
// CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]], float* nonnull align 4 dereferenceable(4) [[REF_TMP15]], float* nonnull align 4 dereferenceable(4) [[REF_TMP16]]) #[[ATTR8]] // CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]], float* nonnull align 4 dereferenceable(4) [[REF_TMP15]], float* nonnull align 4 dereferenceable(4) [[REF_TMP16]]) #[[ATTR9]]
// CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]]) #[[ATTR9]]
// CHECK1-NEXT: [[TMP45:%.*]] = bitcast float* [[REF_TMP16]] to i8* // CHECK1-NEXT: [[TMP45:%.*]] = bitcast float* [[REF_TMP16]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP45]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP45]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP46:%.*]] = bitcast float* [[REF_TMP15]] to i8* // CHECK1-NEXT: [[TMP46:%.*]] = bitcast float* [[REF_TMP15]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP46]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP46]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex"* [[REF_TMP14]] to i8* // CHECK1-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex"* [[REF_TMP14]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP47]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP47]]) #[[ATTR1]]
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue: // CHECK1: omp.body.continue:
Show All 24 Lines
// CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK1-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK1-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8 // CHECK1-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* // CHECK1-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func) // CHECK1-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func)
// CHECK1-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1 // CHECK1-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1
// CHECK1-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]]
// CHECK1: .omp.reduction.then: // CHECK1: .omp.reduction.then:
// CHECK1-NEXT: [[CALL21:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL21:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]]) #[[ATTR9]]
// CHECK1-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]]) // CHECK1-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]])
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DONE]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DONE]]
// CHECK1: .omp.reduction.done: // CHECK1: .omp.reduction.done:
// CHECK1-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8* // CHECK1-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK1-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP63]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP63]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8* // CHECK1-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8*
Show All 21 Lines
// CHECK1-SAME: (%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[__C:%.*]]) #[[ATTR4:[0-9]+]] comdat align 2 { // CHECK1-SAME: (%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[__C:%.*]]) #[[ATTR4:[0-9]+]] comdat align 2 {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex"*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex"*, align 8
// CHECK1-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex"*, align 8 // CHECK1-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex"*, align 8
// CHECK1-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store %"class.std::complex"* [[__C]], %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store %"class.std::complex"* [[__C]], %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: [[TMP0:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[CALL:%.*]] = call float @_ZNKSt7complexIfE4realEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP0]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL:%.*]] = call float @_ZNKSt7complexIfE4realEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP0]]) #[[ATTR9]]
// CHECK1-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP1:%.*]] = load float, float* [[__RE_]], align 4, !tbaa [[TBAA16:![0-9]+]] // CHECK1-NEXT: [[TMP1:%.*]] = load float, float* [[__RE_]], align 4, !tbaa [[TBAA16:![0-9]+]]
// CHECK1-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[CALL]] // CHECK1-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[CALL]]
// CHECK1-NEXT: store float [[ADD]], float* [[__RE_]], align 4, !tbaa [[TBAA16]] // CHECK1-NEXT: store float [[ADD]], float* [[__RE_]], align 4, !tbaa [[TBAA16]]
// CHECK1-NEXT: [[TMP2:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: [[TMP2:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[CALL2:%.*]] = call float @_ZNKSt7complexIfE4imagEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL2:%.*]] = call float @_ZNKSt7complexIfE4imagEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]]) #[[ATTR9]]
// CHECK1-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 1 // CHECK1-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 1
// CHECK1-NEXT: [[TMP3:%.*]] = load float, float* [[__IM_]], align 4, !tbaa [[TBAA18:![0-9]+]] // CHECK1-NEXT: [[TMP3:%.*]] = load float, float* [[__IM_]], align 4, !tbaa [[TBAA18:![0-9]+]]
// CHECK1-NEXT: [[ADD3:%.*]] = fadd float [[TMP3]], [[CALL2]] // CHECK1-NEXT: [[ADD3:%.*]] = fadd float [[TMP3]], [[CALL2]]
// CHECK1-NEXT: store float [[ADD3]], float* [[__IM_]], align 4, !tbaa [[TBAA18]] // CHECK1-NEXT: store float [[ADD3]], float* [[__IM_]], align 4, !tbaa [[TBAA18]]
// CHECK1-NEXT: ret %"class.std::complex"* [[THIS1]] // CHECK1-NEXT: ret %"class.std::complex"* [[THIS1]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func // CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func
// CHECK1-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex", align 4 // CHECK1-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex", align 4
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: br label [[IFCONT6:%.*]] // CHECK1-NEXT: br label [[IFCONT6:%.*]]
// CHECK1: else5: // CHECK1: else5:
// CHECK1-NEXT: br label [[IFCONT6]] // CHECK1-NEXT: br label [[IFCONT6]]
// CHECK1: ifcont6: // CHECK1: ifcont6:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func // CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func
// CHECK1-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() // CHECK1-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x()
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK1-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK1-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: br label [[PRECOND]] // CHECK1-NEXT: br label [[PRECOND]]
// CHECK1: exit: // CHECK1: exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper
// CHECK1-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]] // CHECK1-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]]
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: call void @__kmpc_get_shared_variables(i8*** [[GLOBAL_ARGS]]) // CHECK1-NEXT: call void @__kmpc_get_shared_variables(i8*** [[GLOBAL_ARGS]])
// CHECK1-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[GLOBAL_ARGS]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0
// CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i32** // CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i32**
// CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 1
// CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32** // CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32**
// CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP7]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP7]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 2 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 2
// CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to %"class.std::complex"** // CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to %"class.std::complex"**
// CHECK1-NEXT: [[TMP11:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[TMP10]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: [[TMP11:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[TMP10]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: call void @__omp_outlined__1(i32* [[DOTADDR1]], i32* [[DOTZERO_ADDR]], i32* [[TMP5]], i32* [[TMP8]], %"class.std::complex"* [[TMP11]]) #[[ATTR1]] // CHECK1-NEXT: call void @__omp_outlined__1(i32* [[DOTADDR1]], i32* [[DOTZERO_ADDR]], i32* [[TMP5]], i32* [[TMP8]], %"class.std::complex"* [[TMP11]]) #[[ATTR1]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l19 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16
// CHECK1-SAME: () #[[ATTR0]] { // CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i1 false, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i1 false, i1 true, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[TMP14:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP14:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP14]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP14]]) #[[ATTR1]]
// CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22:![0-9]+]] // CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22:![0-9]+]]
// CHECK1-NEXT: [[TMP15:%.*]] = bitcast double* [[REF_TMP2]] to i8* // CHECK1-NEXT: [[TMP15:%.*]] = bitcast double* [[REF_TMP2]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP15]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP15]]) #[[ATTR1]]
// CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP2]], align 8, !tbaa [[TBAA22]] // CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP2]], align 8, !tbaa [[TBAA22]]
// CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM_ON_STACK]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP2]]) #[[ATTR8]] // CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM_ON_STACK]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP2]]) #[[ATTR9]]
// CHECK1-NEXT: [[TMP16:%.*]] = bitcast double* [[REF_TMP2]] to i8* // CHECK1-NEXT: [[TMP16:%.*]] = bitcast double* [[REF_TMP2]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP16]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP16]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP17:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP17]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP17]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4 // CHECK1-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4
// CHECK1-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[__RE_ADDR:%.*]] = alloca double*, align 8 // CHECK1-NEXT: [[__RE_ADDR:%.*]] = alloca double*, align 8
// CHECK1-NEXT: [[__IM_ADDR:%.*]] = alloca double*, align 8 // CHECK1-NEXT: [[__IM_ADDR:%.*]] = alloca double*, align 8
// CHECK1-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store double* [[__RE]], double** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store double* [[__RE]], double** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store double* [[__IM]], double** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store double* [[__IM]], double** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load double*, double** [[__RE_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load double*, double** [[__RE_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load double*, double** [[__IM_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load double*, double** [[__IM_ADDR]], align 8
// CHECK1-NEXT: call void @_ZNSt7complexIdEC2ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS1]], double* nonnull align 8 dereferenceable(8) [[TMP0]], double* nonnull align 8 dereferenceable(8) [[TMP1]]) #[[ATTR8]] // CHECK1-NEXT: call void @_ZNSt7complexIdEC2ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS1]], double* nonnull align 8 dereferenceable(8) [[TMP0]], double* nonnull align 8 dereferenceable(8) [[TMP1]]) #[[ATTR9]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK1-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 16, i8* [[TMP21]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 16, i8* [[TMP21]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP22:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP22:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP22]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP22]]) #[[ATTR1]]
// CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22]] // CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22]]
// CHECK1-NEXT: [[TMP23:%.*]] = bitcast double* [[REF_TMP6]] to i8* // CHECK1-NEXT: [[TMP23:%.*]] = bitcast double* [[REF_TMP6]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP23]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP23]]) #[[ATTR1]]
// CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP6]], align 8, !tbaa [[TBAA22]] // CHECK1-NEXT: store double 0.000000e+00, double* [[REF_TMP6]], align 8, !tbaa [[TBAA22]]
// CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP6]]) #[[ATTR8]] // CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP6]]) #[[ATTR9]]
// CHECK1-NEXT: [[TMP24:%.*]] = bitcast double* [[REF_TMP6]] to i8* // CHECK1-NEXT: [[TMP24:%.*]] = bitcast double* [[REF_TMP6]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP24]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP24]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK1-NEXT: [[TMP25:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP25]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP25]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8* // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to double // CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to double
// CHECK1-NEXT: store double [[CONV]], double* [[REF_TMP15]], align 8, !tbaa [[TBAA22]] // CHECK1-NEXT: store double [[CONV]], double* [[REF_TMP15]], align 8, !tbaa [[TBAA22]]
// CHECK1-NEXT: [[TMP43:%.*]] = bitcast double* [[REF_TMP16]] to i8* // CHECK1-NEXT: [[TMP43:%.*]] = bitcast double* [[REF_TMP16]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP43]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP43]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to double // CHECK1-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to double
// CHECK1-NEXT: store double [[CONV17]], double* [[REF_TMP16]], align 8, !tbaa [[TBAA22]] // CHECK1-NEXT: store double [[CONV17]], double* [[REF_TMP16]], align 8, !tbaa [[TBAA22]]
// CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]], double* nonnull align 8 dereferenceable(8) [[REF_TMP15]], double* nonnull align 8 dereferenceable(8) [[REF_TMP16]]) #[[ATTR8]] // CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]], double* nonnull align 8 dereferenceable(8) [[REF_TMP15]], double* nonnull align 8 dereferenceable(8) [[REF_TMP16]]) #[[ATTR9]]
// CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]]) #[[ATTR9]]
// CHECK1-NEXT: [[TMP45:%.*]] = bitcast double* [[REF_TMP16]] to i8* // CHECK1-NEXT: [[TMP45:%.*]] = bitcast double* [[REF_TMP16]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP45]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP45]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP46:%.*]] = bitcast double* [[REF_TMP15]] to i8* // CHECK1-NEXT: [[TMP46:%.*]] = bitcast double* [[REF_TMP15]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP46]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP46]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex.0"* [[REF_TMP14]] to i8* // CHECK1-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex.0"* [[REF_TMP14]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP47]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP47]]) #[[ATTR1]]
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue: // CHECK1: omp.body.continue:
Show All 24 Lines
// CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK1-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK1-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8 // CHECK1-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* // CHECK1-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func5, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func6) // CHECK1-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func5, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func6)
// CHECK1-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1 // CHECK1-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1
// CHECK1-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]]
// CHECK1: .omp.reduction.then: // CHECK1: .omp.reduction.then:
// CHECK1-NEXT: [[CALL21:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL21:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]]) #[[ATTR9]]
// CHECK1-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]]) // CHECK1-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]])
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DONE]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DONE]]
// CHECK1: .omp.reduction.done: // CHECK1: .omp.reduction.done:
// CHECK1-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8* // CHECK1-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK1-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP63]]) #[[ATTR1]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP63]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8* // CHECK1-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8*
Show All 21 Lines
// CHECK1-SAME: (%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[__C:%.*]]) #[[ATTR4]] comdat align 2 { // CHECK1-SAME: (%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[__C:%.*]]) #[[ATTR4]] comdat align 2 {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK1-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK1-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK1-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store %"class.std::complex.0"* [[__C]], %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store %"class.std::complex.0"* [[__C]], %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: [[TMP0:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[CALL:%.*]] = call double @_ZNKSt7complexIdE4realEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP0]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL:%.*]] = call double @_ZNKSt7complexIdE4realEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP0]]) #[[ATTR9]]
// CHECK1-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP1:%.*]] = load double, double* [[__RE_]], align 8, !tbaa [[TBAA24:![0-9]+]] // CHECK1-NEXT: [[TMP1:%.*]] = load double, double* [[__RE_]], align 8, !tbaa [[TBAA24:![0-9]+]]
// CHECK1-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], [[CALL]] // CHECK1-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], [[CALL]]
// CHECK1-NEXT: store double [[ADD]], double* [[__RE_]], align 8, !tbaa [[TBAA24]] // CHECK1-NEXT: store double [[ADD]], double* [[__RE_]], align 8, !tbaa [[TBAA24]]
// CHECK1-NEXT: [[TMP2:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: [[TMP2:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[CALL2:%.*]] = call double @_ZNKSt7complexIdE4imagEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]]) #[[ATTR8]] // CHECK1-NEXT: [[CALL2:%.*]] = call double @_ZNKSt7complexIdE4imagEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]]) #[[ATTR9]]
// CHECK1-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1 // CHECK1-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1
// CHECK1-NEXT: [[TMP3:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26:![0-9]+]] // CHECK1-NEXT: [[TMP3:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26:![0-9]+]]
// CHECK1-NEXT: [[ADD3:%.*]] = fadd double [[TMP3]], [[CALL2]] // CHECK1-NEXT: [[ADD3:%.*]] = fadd double [[TMP3]], [[CALL2]]
// CHECK1-NEXT: store double [[ADD3]], double* [[__IM_]], align 8, !tbaa [[TBAA26]] // CHECK1-NEXT: store double [[ADD3]], double* [[__IM_]], align 8, !tbaa [[TBAA26]]
// CHECK1-NEXT: ret %"class.std::complex.0"* [[THIS1]] // CHECK1-NEXT: ret %"class.std::complex.0"* [[THIS1]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func5 // CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func5
// CHECK1-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8 // CHECK1-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex.0", align 8 // CHECK1-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex.0", align 8
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
▲ Show 20 Lines • Show All 74 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: br label [[IFCONT6:%.*]] // CHECK1-NEXT: br label [[IFCONT6:%.*]]
// CHECK1: else5: // CHECK1: else5:
// CHECK1-NEXT: br label [[IFCONT6]] // CHECK1-NEXT: br label [[IFCONT6]]
// CHECK1: ifcont6: // CHECK1: ifcont6:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func6 // CHECK1-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func6
// CHECK1-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() // CHECK1-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x()
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK1-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK1-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: br label [[PRECOND]] // CHECK1-NEXT: br label [[PRECOND]]
// CHECK1: exit: // CHECK1: exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3_wrapper // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3_wrapper
// CHECK1-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]] // CHECK1-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]]
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 93 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK1-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK1-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1 // CHECK1-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1
// CHECK1-NEXT: [[TMP0:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26]] // CHECK1-NEXT: [[TMP0:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26]]
// CHECK1-NEXT: ret double [[TMP0]] // CHECK1-NEXT: ret double [[TMP0]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l19 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16
// CHECK2-SAME: () #[[ATTR0:[0-9]+]] { // CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i1 false, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i1 false, i1 true, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[TMP14:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP14:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP14]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP14]]) #[[ATTR1]]
// CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14:![0-9]+]] // CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14:![0-9]+]]
// CHECK2-NEXT: [[TMP15:%.*]] = bitcast float* [[REF_TMP2]] to i8* // CHECK2-NEXT: [[TMP15:%.*]] = bitcast float* [[REF_TMP2]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP15]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP15]]) #[[ATTR1]]
// CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP2]], align 4, !tbaa [[TBAA14]] // CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP2]], align 4, !tbaa [[TBAA14]]
// CHECK2-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM_ON_STACK]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP2]]) #[[ATTR8:[0-9]+]] // CHECK2-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM_ON_STACK]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP2]]) #[[ATTR9:[0-9]+]]
// CHECK2-NEXT: [[TMP16:%.*]] = bitcast float* [[REF_TMP2]] to i8* // CHECK2-NEXT: [[TMP16:%.*]] = bitcast float* [[REF_TMP2]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP16]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP16]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP17:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP17]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP17]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4 // CHECK2-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4
// CHECK2-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[__RE_ADDR:%.*]] = alloca float*, align 8 // CHECK2-NEXT: [[__RE_ADDR:%.*]] = alloca float*, align 8
// CHECK2-NEXT: [[__IM_ADDR:%.*]] = alloca float*, align 8 // CHECK2-NEXT: [[__IM_ADDR:%.*]] = alloca float*, align 8
// CHECK2-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store float* [[__RE]], float** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store float* [[__RE]], float** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store float* [[__IM]], float** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store float* [[__IM]], float** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load float*, float** [[__RE_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load float*, float** [[__RE_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load float*, float** [[__IM_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load float*, float** [[__IM_ADDR]], align 8
// CHECK2-NEXT: call void @_ZNSt7complexIfEC2ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS1]], float* nonnull align 4 dereferenceable(4) [[TMP0]], float* nonnull align 4 dereferenceable(4) [[TMP1]]) #[[ATTR8]] // CHECK2-NEXT: call void @_ZNSt7complexIfEC2ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS1]], float* nonnull align 4 dereferenceable(4) [[TMP0]], float* nonnull align 4 dereferenceable(4) [[TMP1]]) #[[ATTR9]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK2-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP21]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP21]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP22:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP22:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP22]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP22]]) #[[ATTR1]]
// CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14]] // CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14]]
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast float* [[REF_TMP6]] to i8* // CHECK2-NEXT: [[TMP23:%.*]] = bitcast float* [[REF_TMP6]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP23]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP23]]) #[[ATTR1]]
// CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP6]], align 4, !tbaa [[TBAA14]] // CHECK2-NEXT: store float 0.000000e+00, float* [[REF_TMP6]], align 4, !tbaa [[TBAA14]]
// CHECK2-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP6]]) #[[ATTR8]] // CHECK2-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP6]]) #[[ATTR9]]
// CHECK2-NEXT: [[TMP24:%.*]] = bitcast float* [[REF_TMP6]] to i8* // CHECK2-NEXT: [[TMP24:%.*]] = bitcast float* [[REF_TMP6]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP24]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP24]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP25:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP25]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP25]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8* // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to float // CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to float
// CHECK2-NEXT: store float [[CONV]], float* [[REF_TMP15]], align 4, !tbaa [[TBAA14]] // CHECK2-NEXT: store float [[CONV]], float* [[REF_TMP15]], align 4, !tbaa [[TBAA14]]
// CHECK2-NEXT: [[TMP43:%.*]] = bitcast float* [[REF_TMP16]] to i8* // CHECK2-NEXT: [[TMP43:%.*]] = bitcast float* [[REF_TMP16]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP43]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP43]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to float // CHECK2-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to float
// CHECK2-NEXT: store float [[CONV17]], float* [[REF_TMP16]], align 4, !tbaa [[TBAA14]] // CHECK2-NEXT: store float [[CONV17]], float* [[REF_TMP16]], align 4, !tbaa [[TBAA14]]
// CHECK2-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]], float* nonnull align 4 dereferenceable(4) [[REF_TMP15]], float* nonnull align 4 dereferenceable(4) [[REF_TMP16]]) #[[ATTR8]] // CHECK2-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]], float* nonnull align 4 dereferenceable(4) [[REF_TMP15]], float* nonnull align 4 dereferenceable(4) [[REF_TMP16]]) #[[ATTR9]]
// CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]]) #[[ATTR9]]
// CHECK2-NEXT: [[TMP45:%.*]] = bitcast float* [[REF_TMP16]] to i8* // CHECK2-NEXT: [[TMP45:%.*]] = bitcast float* [[REF_TMP16]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP45]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP45]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP46:%.*]] = bitcast float* [[REF_TMP15]] to i8* // CHECK2-NEXT: [[TMP46:%.*]] = bitcast float* [[REF_TMP15]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP46]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP46]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex"* [[REF_TMP14]] to i8* // CHECK2-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex"* [[REF_TMP14]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP47]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP47]]) #[[ATTR1]]
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue: // CHECK2: omp.body.continue:
Show All 24 Lines
// CHECK2-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK2-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK2-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK2-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8 // CHECK2-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8
// CHECK2-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* // CHECK2-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func) // CHECK2-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func)
// CHECK2-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1 // CHECK2-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1
// CHECK2-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]] // CHECK2-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]]
// CHECK2: .omp.reduction.then: // CHECK2: .omp.reduction.then:
// CHECK2-NEXT: [[CALL21:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL21:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]]) #[[ATTR9]]
// CHECK2-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]]) // CHECK2-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]])
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DONE]] // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DONE]]
// CHECK2: .omp.reduction.done: // CHECK2: .omp.reduction.done:
// CHECK2-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8* // CHECK2-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK2-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP63]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP63]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8* // CHECK2-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8*
Show All 21 Lines
// CHECK2-SAME: (%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[__C:%.*]]) #[[ATTR4:[0-9]+]] comdat align 2 { // CHECK2-SAME: (%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[__C:%.*]]) #[[ATTR4:[0-9]+]] comdat align 2 {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex"*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex"*, align 8
// CHECK2-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex"*, align 8 // CHECK2-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex"*, align 8
// CHECK2-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store %"class.std::complex"* [[__C]], %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store %"class.std::complex"* [[__C]], %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: [[TMP0:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[CALL:%.*]] = call float @_ZNKSt7complexIfE4realEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP0]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL:%.*]] = call float @_ZNKSt7complexIfE4realEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP0]]) #[[ATTR9]]
// CHECK2-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 0 // CHECK2-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT: [[TMP1:%.*]] = load float, float* [[__RE_]], align 4, !tbaa [[TBAA16:![0-9]+]] // CHECK2-NEXT: [[TMP1:%.*]] = load float, float* [[__RE_]], align 4, !tbaa [[TBAA16:![0-9]+]]
// CHECK2-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[CALL]] // CHECK2-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[CALL]]
// CHECK2-NEXT: store float [[ADD]], float* [[__RE_]], align 4, !tbaa [[TBAA16]] // CHECK2-NEXT: store float [[ADD]], float* [[__RE_]], align 4, !tbaa [[TBAA16]]
// CHECK2-NEXT: [[TMP2:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: [[TMP2:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[CALL2:%.*]] = call float @_ZNKSt7complexIfE4imagEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL2:%.*]] = call float @_ZNKSt7complexIfE4imagEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]]) #[[ATTR9]]
// CHECK2-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 1 // CHECK2-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 1
// CHECK2-NEXT: [[TMP3:%.*]] = load float, float* [[__IM_]], align 4, !tbaa [[TBAA18:![0-9]+]] // CHECK2-NEXT: [[TMP3:%.*]] = load float, float* [[__IM_]], align 4, !tbaa [[TBAA18:![0-9]+]]
// CHECK2-NEXT: [[ADD3:%.*]] = fadd float [[TMP3]], [[CALL2]] // CHECK2-NEXT: [[ADD3:%.*]] = fadd float [[TMP3]], [[CALL2]]
// CHECK2-NEXT: store float [[ADD3]], float* [[__IM_]], align 4, !tbaa [[TBAA18]] // CHECK2-NEXT: store float [[ADD3]], float* [[__IM_]], align 4, !tbaa [[TBAA18]]
// CHECK2-NEXT: ret %"class.std::complex"* [[THIS1]] // CHECK2-NEXT: ret %"class.std::complex"* [[THIS1]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func // CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func
// CHECK2-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex", align 4 // CHECK2-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex", align 4
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: br label [[IFCONT6:%.*]] // CHECK2-NEXT: br label [[IFCONT6:%.*]]
// CHECK2: else5: // CHECK2: else5:
// CHECK2-NEXT: br label [[IFCONT6]] // CHECK2-NEXT: br label [[IFCONT6]]
// CHECK2: ifcont6: // CHECK2: ifcont6:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func // CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func
// CHECK2-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() // CHECK2-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x()
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK2-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK2-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: br label [[PRECOND]] // CHECK2-NEXT: br label [[PRECOND]]
// CHECK2: exit: // CHECK2: exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper
// CHECK2-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8 // CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]] // CHECK2-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]]
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: call void @__kmpc_get_shared_variables(i8*** [[GLOBAL_ARGS]]) // CHECK2-NEXT: call void @__kmpc_get_shared_variables(i8*** [[GLOBAL_ARGS]])
// CHECK2-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[GLOBAL_ARGS]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[GLOBAL_ARGS]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0
// CHECK2-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i32** // CHECK2-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i32**
// CHECK2-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 1 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 1
// CHECK2-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32** // CHECK2-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32**
// CHECK2-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP7]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP7]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 2 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 2
// CHECK2-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to %"class.std::complex"** // CHECK2-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to %"class.std::complex"**
// CHECK2-NEXT: [[TMP11:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[TMP10]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: [[TMP11:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[TMP10]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: call void @__omp_outlined__1(i32* [[DOTADDR1]], i32* [[DOTZERO_ADDR]], i32* [[TMP5]], i32* [[TMP8]], %"class.std::complex"* [[TMP11]]) #[[ATTR1]] // CHECK2-NEXT: call void @__omp_outlined__1(i32* [[DOTADDR1]], i32* [[DOTZERO_ADDR]], i32* [[TMP5]], i32* [[TMP8]], %"class.std::complex"* [[TMP11]]) #[[ATTR1]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l19 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16
// CHECK2-SAME: () #[[ATTR0]] { // CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i1 false, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i1 false, i1 true, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[TMP14:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP14:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP14]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP14]]) #[[ATTR1]]
// CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22:![0-9]+]] // CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22:![0-9]+]]
// CHECK2-NEXT: [[TMP15:%.*]] = bitcast double* [[REF_TMP2]] to i8* // CHECK2-NEXT: [[TMP15:%.*]] = bitcast double* [[REF_TMP2]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP15]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP15]]) #[[ATTR1]]
// CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP2]], align 8, !tbaa [[TBAA22]] // CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP2]], align 8, !tbaa [[TBAA22]]
// CHECK2-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM_ON_STACK]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP2]]) #[[ATTR8]] // CHECK2-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM_ON_STACK]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP2]]) #[[ATTR9]]
// CHECK2-NEXT: [[TMP16:%.*]] = bitcast double* [[REF_TMP2]] to i8* // CHECK2-NEXT: [[TMP16:%.*]] = bitcast double* [[REF_TMP2]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP16]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP16]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP17:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP17]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP17]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4 // CHECK2-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4
// CHECK2-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[__RE_ADDR:%.*]] = alloca double*, align 8 // CHECK2-NEXT: [[__RE_ADDR:%.*]] = alloca double*, align 8
// CHECK2-NEXT: [[__IM_ADDR:%.*]] = alloca double*, align 8 // CHECK2-NEXT: [[__IM_ADDR:%.*]] = alloca double*, align 8
// CHECK2-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store double* [[__RE]], double** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store double* [[__RE]], double** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store double* [[__IM]], double** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store double* [[__IM]], double** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load double*, double** [[__RE_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load double*, double** [[__RE_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load double*, double** [[__IM_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load double*, double** [[__IM_ADDR]], align 8
// CHECK2-NEXT: call void @_ZNSt7complexIdEC2ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS1]], double* nonnull align 8 dereferenceable(8) [[TMP0]], double* nonnull align 8 dereferenceable(8) [[TMP1]]) #[[ATTR8]] // CHECK2-NEXT: call void @_ZNSt7complexIdEC2ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS1]], double* nonnull align 8 dereferenceable(8) [[TMP0]], double* nonnull align 8 dereferenceable(8) [[TMP1]]) #[[ATTR9]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK2-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 16, i8* [[TMP21]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 16, i8* [[TMP21]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP22:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP22:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP22]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP22]]) #[[ATTR1]]
// CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22]] // CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22]]
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast double* [[REF_TMP6]] to i8* // CHECK2-NEXT: [[TMP23:%.*]] = bitcast double* [[REF_TMP6]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP23]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP23]]) #[[ATTR1]]
// CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP6]], align 8, !tbaa [[TBAA22]] // CHECK2-NEXT: store double 0.000000e+00, double* [[REF_TMP6]], align 8, !tbaa [[TBAA22]]
// CHECK2-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP6]]) #[[ATTR8]] // CHECK2-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP6]]) #[[ATTR9]]
// CHECK2-NEXT: [[TMP24:%.*]] = bitcast double* [[REF_TMP6]] to i8* // CHECK2-NEXT: [[TMP24:%.*]] = bitcast double* [[REF_TMP6]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP24]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP24]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK2-NEXT: [[TMP25:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP25]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP25]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8* // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to double // CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to double
// CHECK2-NEXT: store double [[CONV]], double* [[REF_TMP15]], align 8, !tbaa [[TBAA22]] // CHECK2-NEXT: store double [[CONV]], double* [[REF_TMP15]], align 8, !tbaa [[TBAA22]]
// CHECK2-NEXT: [[TMP43:%.*]] = bitcast double* [[REF_TMP16]] to i8* // CHECK2-NEXT: [[TMP43:%.*]] = bitcast double* [[REF_TMP16]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP43]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP43]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to double // CHECK2-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to double
// CHECK2-NEXT: store double [[CONV17]], double* [[REF_TMP16]], align 8, !tbaa [[TBAA22]] // CHECK2-NEXT: store double [[CONV17]], double* [[REF_TMP16]], align 8, !tbaa [[TBAA22]]
// CHECK2-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]], double* nonnull align 8 dereferenceable(8) [[REF_TMP15]], double* nonnull align 8 dereferenceable(8) [[REF_TMP16]]) #[[ATTR8]] // CHECK2-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]], double* nonnull align 8 dereferenceable(8) [[REF_TMP15]], double* nonnull align 8 dereferenceable(8) [[REF_TMP16]]) #[[ATTR9]]
// CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]]) #[[ATTR9]]
// CHECK2-NEXT: [[TMP45:%.*]] = bitcast double* [[REF_TMP16]] to i8* // CHECK2-NEXT: [[TMP45:%.*]] = bitcast double* [[REF_TMP16]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP45]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP45]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP46:%.*]] = bitcast double* [[REF_TMP15]] to i8* // CHECK2-NEXT: [[TMP46:%.*]] = bitcast double* [[REF_TMP15]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP46]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP46]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex.0"* [[REF_TMP14]] to i8* // CHECK2-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex.0"* [[REF_TMP14]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP47]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP47]]) #[[ATTR1]]
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue: // CHECK2: omp.body.continue:
Show All 24 Lines
// CHECK2-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK2-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK2-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK2-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8 // CHECK2-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8
// CHECK2-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* // CHECK2-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func5, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func6) // CHECK2-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func5, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func6)
// CHECK2-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1 // CHECK2-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1
// CHECK2-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]] // CHECK2-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]]
// CHECK2: .omp.reduction.then: // CHECK2: .omp.reduction.then:
// CHECK2-NEXT: [[CALL21:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL21:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]]) #[[ATTR9]]
// CHECK2-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]]) // CHECK2-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]])
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DONE]] // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DONE]]
// CHECK2: .omp.reduction.done: // CHECK2: .omp.reduction.done:
// CHECK2-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8* // CHECK2-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK2-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP63]]) #[[ATTR1]] // CHECK2-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP63]]) #[[ATTR1]]
// CHECK2-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8* // CHECK2-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8*
Show All 21 Lines
// CHECK2-SAME: (%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[__C:%.*]]) #[[ATTR4]] comdat align 2 { // CHECK2-SAME: (%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[__C:%.*]]) #[[ATTR4]] comdat align 2 {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK2-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK2-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK2-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store %"class.std::complex.0"* [[__C]], %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store %"class.std::complex.0"* [[__C]], %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: [[TMP0:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[CALL:%.*]] = call double @_ZNKSt7complexIdE4realEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP0]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL:%.*]] = call double @_ZNKSt7complexIdE4realEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP0]]) #[[ATTR9]]
// CHECK2-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 0 // CHECK2-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT: [[TMP1:%.*]] = load double, double* [[__RE_]], align 8, !tbaa [[TBAA24:![0-9]+]] // CHECK2-NEXT: [[TMP1:%.*]] = load double, double* [[__RE_]], align 8, !tbaa [[TBAA24:![0-9]+]]
// CHECK2-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], [[CALL]] // CHECK2-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], [[CALL]]
// CHECK2-NEXT: store double [[ADD]], double* [[__RE_]], align 8, !tbaa [[TBAA24]] // CHECK2-NEXT: store double [[ADD]], double* [[__RE_]], align 8, !tbaa [[TBAA24]]
// CHECK2-NEXT: [[TMP2:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: [[TMP2:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[CALL2:%.*]] = call double @_ZNKSt7complexIdE4imagEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]]) #[[ATTR8]] // CHECK2-NEXT: [[CALL2:%.*]] = call double @_ZNKSt7complexIdE4imagEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]]) #[[ATTR9]]
// CHECK2-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1 // CHECK2-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1
// CHECK2-NEXT: [[TMP3:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26:![0-9]+]] // CHECK2-NEXT: [[TMP3:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26:![0-9]+]]
// CHECK2-NEXT: [[ADD3:%.*]] = fadd double [[TMP3]], [[CALL2]] // CHECK2-NEXT: [[ADD3:%.*]] = fadd double [[TMP3]], [[CALL2]]
// CHECK2-NEXT: store double [[ADD3]], double* [[__IM_]], align 8, !tbaa [[TBAA26]] // CHECK2-NEXT: store double [[ADD3]], double* [[__IM_]], align 8, !tbaa [[TBAA26]]
// CHECK2-NEXT: ret %"class.std::complex.0"* [[THIS1]] // CHECK2-NEXT: ret %"class.std::complex.0"* [[THIS1]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func5 // CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func5
// CHECK2-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR5]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8 // CHECK2-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex.0", align 8 // CHECK2-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex.0", align 8
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
▲ Show 20 Lines • Show All 74 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: br label [[IFCONT6:%.*]] // CHECK2-NEXT: br label [[IFCONT6:%.*]]
// CHECK2: else5: // CHECK2: else5:
// CHECK2-NEXT: br label [[IFCONT6]] // CHECK2-NEXT: br label [[IFCONT6]]
// CHECK2: ifcont6: // CHECK2: ifcont6:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func6 // CHECK2-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func6
// CHECK2-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() // CHECK2-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x()
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK2-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK2-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]]
// CHECK2-NEXT: br label [[PRECOND]] // CHECK2-NEXT: br label [[PRECOND]]
// CHECK2: exit: // CHECK2: exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3_wrapper // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3_wrapper
// CHECK2-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8 // CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]] // CHECK2-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]]
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK2-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 93 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK2-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK2-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1 // CHECK2-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1
// CHECK2-NEXT: [[TMP0:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26]] // CHECK2-NEXT: [[TMP0:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26]]
// CHECK2-NEXT: ret double [[TMP0]] // CHECK2-NEXT: ret double [[TMP0]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l19 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] { // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i1 false, i1 true, i1 true) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i1 false, i1 true, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[TMP14:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP14:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP14]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP14]]) #[[ATTR1]]
// CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14:![0-9]+]] // CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14:![0-9]+]]
// CHECK3-NEXT: [[TMP15:%.*]] = bitcast float* [[REF_TMP2]] to i8* // CHECK3-NEXT: [[TMP15:%.*]] = bitcast float* [[REF_TMP2]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP15]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP15]]) #[[ATTR1]]
// CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP2]], align 4, !tbaa [[TBAA14]] // CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP2]], align 4, !tbaa [[TBAA14]]
// CHECK3-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM_ON_STACK]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP2]]) #[[ATTR8:[0-9]+]] // CHECK3-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM_ON_STACK]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP2]]) #[[ATTR9:[0-9]+]]
// CHECK3-NEXT: [[TMP16:%.*]] = bitcast float* [[REF_TMP2]] to i8* // CHECK3-NEXT: [[TMP16:%.*]] = bitcast float* [[REF_TMP2]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP16]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP16]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP17:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP17]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP17]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4 // CHECK3-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4
// CHECK3-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[__RE_ADDR:%.*]] = alloca float*, align 8 // CHECK3-NEXT: [[__RE_ADDR:%.*]] = alloca float*, align 8
// CHECK3-NEXT: [[__IM_ADDR:%.*]] = alloca float*, align 8 // CHECK3-NEXT: [[__IM_ADDR:%.*]] = alloca float*, align 8
// CHECK3-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store float* [[__RE]], float** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store float* [[__RE]], float** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store float* [[__IM]], float** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store float* [[__IM]], float** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8 // CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load float*, float** [[__RE_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load float*, float** [[__RE_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load float*, float** [[__IM_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load float*, float** [[__IM_ADDR]], align 8
// CHECK3-NEXT: call void @_ZNSt7complexIfEC2ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS1]], float* nonnull align 4 dereferenceable(4) [[TMP0]], float* nonnull align 4 dereferenceable(4) [[TMP1]]) #[[ATTR8]] // CHECK3-NEXT: call void @_ZNSt7complexIfEC2ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS1]], float* nonnull align 4 dereferenceable(4) [[TMP0]], float* nonnull align 4 dereferenceable(4) [[TMP1]]) #[[ATTR9]]
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK3-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP21]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP21]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP22:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP22:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP22]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP22]]) #[[ATTR1]]
// CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14]] // CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP]], align 4, !tbaa [[TBAA14]]
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast float* [[REF_TMP6]] to i8* // CHECK3-NEXT: [[TMP23:%.*]] = bitcast float* [[REF_TMP6]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP23]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP23]]) #[[ATTR1]]
// CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP6]], align 4, !tbaa [[TBAA14]] // CHECK3-NEXT: store float 0.000000e+00, float* [[REF_TMP6]], align 4, !tbaa [[TBAA14]]
// CHECK3-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP6]]) #[[ATTR8]] // CHECK3-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], float* nonnull align 4 dereferenceable(4) [[REF_TMP]], float* nonnull align 4 dereferenceable(4) [[REF_TMP6]]) #[[ATTR9]]
// CHECK3-NEXT: [[TMP24:%.*]] = bitcast float* [[REF_TMP6]] to i8* // CHECK3-NEXT: [[TMP24:%.*]] = bitcast float* [[REF_TMP6]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP24]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP24]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast float* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP25:%.*]] = bitcast float* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP25]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP25]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8* // CHECK3-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK3-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to float // CHECK3-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to float
// CHECK3-NEXT: store float [[CONV]], float* [[REF_TMP15]], align 4, !tbaa [[TBAA14]] // CHECK3-NEXT: store float [[CONV]], float* [[REF_TMP15]], align 4, !tbaa [[TBAA14]]
// CHECK3-NEXT: [[TMP43:%.*]] = bitcast float* [[REF_TMP16]] to i8* // CHECK3-NEXT: [[TMP43:%.*]] = bitcast float* [[REF_TMP16]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP43]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP43]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to float // CHECK3-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to float
// CHECK3-NEXT: store float [[CONV17]], float* [[REF_TMP16]], align 4, !tbaa [[TBAA14]] // CHECK3-NEXT: store float [[CONV17]], float* [[REF_TMP16]], align 4, !tbaa [[TBAA14]]
// CHECK3-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]], float* nonnull align 4 dereferenceable(4) [[REF_TMP15]], float* nonnull align 4 dereferenceable(4) [[REF_TMP16]]) #[[ATTR8]] // CHECK3-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]], float* nonnull align 4 dereferenceable(4) [[REF_TMP15]], float* nonnull align 4 dereferenceable(4) [[REF_TMP16]]) #[[ATTR9]]
// CHECK3-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[REF_TMP14]]) #[[ATTR9]]
// CHECK3-NEXT: [[TMP45:%.*]] = bitcast float* [[REF_TMP16]] to i8* // CHECK3-NEXT: [[TMP45:%.*]] = bitcast float* [[REF_TMP16]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP45]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP45]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP46:%.*]] = bitcast float* [[REF_TMP15]] to i8* // CHECK3-NEXT: [[TMP46:%.*]] = bitcast float* [[REF_TMP15]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP46]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP46]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex"* [[REF_TMP14]] to i8* // CHECK3-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex"* [[REF_TMP14]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP47]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP47]]) #[[ATTR1]]
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue: // CHECK3: omp.body.continue:
Show All 24 Lines
// CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK3-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK3-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK3-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8 // CHECK3-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8
// CHECK3-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* // CHECK3-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK3-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func) // CHECK3-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func)
// CHECK3-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1 // CHECK3-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1
// CHECK3-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]] // CHECK3-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]]
// CHECK3: .omp.reduction.then: // CHECK3: .omp.reduction.then:
// CHECK3-NEXT: [[CALL21:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL21:%.*]] = call nonnull align 4 dereferenceable(8) %"class.std::complex"* @_ZNSt7complexIfEpLIfEERS0_RKS_IT_E(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[PARTIAL_SUM5]]) #[[ATTR9]]
// CHECK3-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]]) // CHECK3-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]])
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DONE]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DONE]]
// CHECK3: .omp.reduction.done: // CHECK3: .omp.reduction.done:
// CHECK3-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8* // CHECK3-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8* // CHECK3-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex"* [[PARTIAL_SUM5]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP63]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP63]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8* // CHECK3-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8*
Show All 21 Lines
// CHECK3-SAME: (%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[__C:%.*]]) #[[ATTR4:[0-9]+]] comdat align 2 { // CHECK3-SAME: (%"class.std::complex"* nonnull align 4 dereferenceable(8) [[THIS:%.*]], %"class.std::complex"* nonnull align 4 dereferenceable(8) [[__C:%.*]]) #[[ATTR4:[0-9]+]] comdat align 2 {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex"*, align 8 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex"*, align 8
// CHECK3-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex"*, align 8 // CHECK3-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex"*, align 8
// CHECK3-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store %"class.std::complex"* [[THIS]], %"class.std::complex"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store %"class.std::complex"* [[__C]], %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store %"class.std::complex"* [[__C]], %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8 // CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: [[TMP0:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[CALL:%.*]] = call float @_ZNKSt7complexIfE4realEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP0]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL:%.*]] = call float @_ZNKSt7complexIfE4realEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP0]]) #[[ATTR9]]
// CHECK3-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP1:%.*]] = load float, float* [[__RE_]], align 4, !tbaa [[TBAA16:![0-9]+]] // CHECK3-NEXT: [[TMP1:%.*]] = load float, float* [[__RE_]], align 4, !tbaa [[TBAA16:![0-9]+]]
// CHECK3-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[CALL]] // CHECK3-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[CALL]]
// CHECK3-NEXT: store float [[ADD]], float* [[__RE_]], align 4, !tbaa [[TBAA16]] // CHECK3-NEXT: store float [[ADD]], float* [[__RE_]], align 4, !tbaa [[TBAA16]]
// CHECK3-NEXT: [[TMP2:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: [[TMP2:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[CALL2:%.*]] = call float @_ZNKSt7complexIfE4imagEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL2:%.*]] = call float @_ZNKSt7complexIfE4imagEv(%"class.std::complex"* nonnull align 4 dereferenceable(8) [[TMP2]]) #[[ATTR9]]
// CHECK3-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 1 // CHECK3-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex", %"class.std::complex"* [[THIS1]], i32 0, i32 1
// CHECK3-NEXT: [[TMP3:%.*]] = load float, float* [[__IM_]], align 4, !tbaa [[TBAA18:![0-9]+]] // CHECK3-NEXT: [[TMP3:%.*]] = load float, float* [[__IM_]], align 4, !tbaa [[TBAA18:![0-9]+]]
// CHECK3-NEXT: [[ADD3:%.*]] = fadd float [[TMP3]], [[CALL2]] // CHECK3-NEXT: [[ADD3:%.*]] = fadd float [[TMP3]], [[CALL2]]
// CHECK3-NEXT: store float [[ADD3]], float* [[__IM_]], align 4, !tbaa [[TBAA18]] // CHECK3-NEXT: store float [[ADD3]], float* [[__IM_]], align 4, !tbaa [[TBAA18]]
// CHECK3-NEXT: ret %"class.std::complex"* [[THIS1]] // CHECK3-NEXT: ret %"class.std::complex"* [[THIS1]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func // CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func
// CHECK3-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8 // CHECK3-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK3-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex", align 4 // CHECK3-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex", align 4
// CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: br label [[IFCONT6:%.*]] // CHECK3-NEXT: br label [[IFCONT6:%.*]]
// CHECK3: else5: // CHECK3: else5:
// CHECK3-NEXT: br label [[IFCONT6]] // CHECK3-NEXT: br label [[IFCONT6]]
// CHECK3: ifcont6: // CHECK3: ifcont6:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func // CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func
// CHECK3-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
// CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() // CHECK3-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x()
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK3-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK3-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: br label [[PRECOND]] // CHECK3-NEXT: br label [[PRECOND]]
// CHECK3: exit: // CHECK3: exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper
// CHECK3-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8 // CHECK3-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8
// CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]] // CHECK3-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]]
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: call void @__kmpc_get_shared_variables(i8*** [[GLOBAL_ARGS]]) // CHECK3-NEXT: call void @__kmpc_get_shared_variables(i8*** [[GLOBAL_ARGS]])
// CHECK3-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[GLOBAL_ARGS]], align 8 // CHECK3-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[GLOBAL_ARGS]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0
// CHECK3-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i32** // CHECK3-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i32**
// CHECK3-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 1
// CHECK3-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32** // CHECK3-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32**
// CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP7]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP7]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 2 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 2
// CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to %"class.std::complex"** // CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to %"class.std::complex"**
// CHECK3-NEXT: [[TMP11:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[TMP10]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: [[TMP11:%.*]] = load %"class.std::complex"*, %"class.std::complex"** [[TMP10]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: call void @__omp_outlined__1(i32* [[DOTADDR1]], i32* [[DOTZERO_ADDR]], i32* [[TMP5]], i32* [[TMP8]], %"class.std::complex"* [[TMP11]]) #[[ATTR1]] // CHECK3-NEXT: call void @__omp_outlined__1(i32* [[DOTADDR1]], i32* [[DOTZERO_ADDR]], i32* [[TMP5]], i32* [[TMP8]], %"class.std::complex"* [[TMP11]]) #[[ATTR1]]
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l19 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16
// CHECK3-SAME: () #[[ATTR0]] { // CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i1 false, i1 true, i1 true) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i1 false, i1 true, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[ADD]], i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[TMP14:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP14:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP14]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP14]]) #[[ATTR1]]
// CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22:![0-9]+]] // CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22:![0-9]+]]
// CHECK3-NEXT: [[TMP15:%.*]] = bitcast double* [[REF_TMP2]] to i8* // CHECK3-NEXT: [[TMP15:%.*]] = bitcast double* [[REF_TMP2]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP15]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP15]]) #[[ATTR1]]
// CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP2]], align 8, !tbaa [[TBAA22]] // CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP2]], align 8, !tbaa [[TBAA22]]
// CHECK3-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM_ON_STACK]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP2]]) #[[ATTR8]] // CHECK3-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM_ON_STACK]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP2]]) #[[ATTR9]]
// CHECK3-NEXT: [[TMP16:%.*]] = bitcast double* [[REF_TMP2]] to i8* // CHECK3-NEXT: [[TMP16:%.*]] = bitcast double* [[REF_TMP2]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP16]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP16]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP17:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP17]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP17]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4 // CHECK3-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP18]], 4
// CHECK3-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[MUL3]], i32* [[ISTART_ON_STACK]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[IB]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[__RE_ADDR:%.*]] = alloca double*, align 8 // CHECK3-NEXT: [[__RE_ADDR:%.*]] = alloca double*, align 8
// CHECK3-NEXT: [[__IM_ADDR:%.*]] = alloca double*, align 8 // CHECK3-NEXT: [[__IM_ADDR:%.*]] = alloca double*, align 8
// CHECK3-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store double* [[__RE]], double** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store double* [[__RE]], double** [[__RE_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store double* [[__IM]], double** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store double* [[__IM]], double** [[__IM_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load double*, double** [[__RE_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load double*, double** [[__RE_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load double*, double** [[__IM_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load double*, double** [[__IM_ADDR]], align 8
// CHECK3-NEXT: call void @_ZNSt7complexIdEC2ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS1]], double* nonnull align 8 dereferenceable(8) [[TMP0]], double* nonnull align 8 dereferenceable(8) [[TMP1]]) #[[ATTR8]] // CHECK3-NEXT: call void @_ZNSt7complexIdEC2ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS1]], double* nonnull align 8 dereferenceable(8) [[TMP0]], double* nonnull align 8 dereferenceable(8) [[TMP1]]) #[[ATTR9]]
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__3 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__3
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ISTART:%.*]], i32* nonnull align 4 dereferenceable(4) [[IEND:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK3-NEXT: [[TMP21:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 16, i8* [[TMP21]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 16, i8* [[TMP21]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP22:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP22:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP22]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP22]]) #[[ATTR1]]
// CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22]] // CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP]], align 8, !tbaa [[TBAA22]]
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast double* [[REF_TMP6]] to i8* // CHECK3-NEXT: [[TMP23:%.*]] = bitcast double* [[REF_TMP6]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP23]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP23]]) #[[ATTR1]]
// CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP6]], align 8, !tbaa [[TBAA22]] // CHECK3-NEXT: store double 0.000000e+00, double* [[REF_TMP6]], align 8, !tbaa [[TBAA22]]
// CHECK3-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP6]]) #[[ATTR8]] // CHECK3-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], double* nonnull align 8 dereferenceable(8) [[REF_TMP]], double* nonnull align 8 dereferenceable(8) [[REF_TMP6]]) #[[ATTR9]]
// CHECK3-NEXT: [[TMP24:%.*]] = bitcast double* [[REF_TMP6]] to i8* // CHECK3-NEXT: [[TMP24:%.*]] = bitcast double* [[REF_TMP6]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP24]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP24]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast double* [[REF_TMP]] to i8* // CHECK3-NEXT: [[TMP25:%.*]] = bitcast double* [[REF_TMP]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP25]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP25]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8* // CHECK3-NEXT: [[TMP26:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* [[TMP26]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK3-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP42:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to double // CHECK3-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP42]] to double
// CHECK3-NEXT: store double [[CONV]], double* [[REF_TMP15]], align 8, !tbaa [[TBAA22]] // CHECK3-NEXT: store double [[CONV]], double* [[REF_TMP15]], align 8, !tbaa [[TBAA22]]
// CHECK3-NEXT: [[TMP43:%.*]] = bitcast double* [[REF_TMP16]] to i8* // CHECK3-NEXT: [[TMP43:%.*]] = bitcast double* [[REF_TMP16]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP43]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[TMP43]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: [[TMP44:%.*]] = load i32, i32* [[I7]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to double // CHECK3-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP44]] to double
// CHECK3-NEXT: store double [[CONV17]], double* [[REF_TMP16]], align 8, !tbaa [[TBAA22]] // CHECK3-NEXT: store double [[CONV17]], double* [[REF_TMP16]], align 8, !tbaa [[TBAA22]]
// CHECK3-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]], double* nonnull align 8 dereferenceable(8) [[REF_TMP15]], double* nonnull align 8 dereferenceable(8) [[REF_TMP16]]) #[[ATTR8]] // CHECK3-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]], double* nonnull align 8 dereferenceable(8) [[REF_TMP15]], double* nonnull align 8 dereferenceable(8) [[REF_TMP16]]) #[[ATTR9]]
// CHECK3-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[REF_TMP14]]) #[[ATTR9]]
// CHECK3-NEXT: [[TMP45:%.*]] = bitcast double* [[REF_TMP16]] to i8* // CHECK3-NEXT: [[TMP45:%.*]] = bitcast double* [[REF_TMP16]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP45]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP45]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP46:%.*]] = bitcast double* [[REF_TMP15]] to i8* // CHECK3-NEXT: [[TMP46:%.*]] = bitcast double* [[REF_TMP15]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP46]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[TMP46]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex.0"* [[REF_TMP14]] to i8* // CHECK3-NEXT: [[TMP47:%.*]] = bitcast %"class.std::complex.0"* [[REF_TMP14]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP47]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP47]]) #[[ATTR1]]
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue: // CHECK3: omp.body.continue:
Show All 24 Lines
// CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK3-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK3-NEXT: [[TMP58:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK3-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8 // CHECK3-NEXT: store i8* [[TMP58]], i8** [[TMP57]], align 8
// CHECK3-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* // CHECK3-NEXT: [[TMP59:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK3-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func5, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func6) // CHECK3-NEXT: [[TMP60:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i64 8, i8* [[TMP59]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func5, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func6)
// CHECK3-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1 // CHECK3-NEXT: [[TMP61:%.*]] = icmp eq i32 [[TMP60]], 1
// CHECK3-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]] // CHECK3-NEXT: br i1 [[TMP61]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]]
// CHECK3: .omp.reduction.then: // CHECK3: .omp.reduction.then:
// CHECK3-NEXT: [[CALL21:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL21:%.*]] = call nonnull align 8 dereferenceable(16) %"class.std::complex.0"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[PARTIAL_SUM5]]) #[[ATTR9]]
// CHECK3-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]]) // CHECK3-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP56]])
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DONE]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DONE]]
// CHECK3: .omp.reduction.done: // CHECK3: .omp.reduction.done:
// CHECK3-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8* // CHECK3-NEXT: [[TMP62:%.*]] = bitcast i32* [[I7]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 4, i8* [[TMP62]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8* // CHECK3-NEXT: [[TMP63:%.*]] = bitcast %"class.std::complex.0"* [[PARTIAL_SUM5]] to i8*
// CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP63]]) #[[ATTR1]] // CHECK3-NEXT: call void @llvm.lifetime.end.p0i8(i64 16, i8* [[TMP63]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8* // CHECK3-NEXT: [[TMP64:%.*]] = bitcast i32* [[DOTOMP_IS_LAST]] to i8*
Show All 21 Lines
// CHECK3-SAME: (%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[__C:%.*]]) #[[ATTR4]] comdat align 2 { // CHECK3-SAME: (%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[THIS:%.*]], %"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[__C:%.*]]) #[[ATTR4]] comdat align 2 {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK3-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8 // CHECK3-NEXT: [[__C_ADDR:%.*]] = alloca %"class.std::complex.0"*, align 8
// CHECK3-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store %"class.std::complex.0"* [[THIS]], %"class.std::complex.0"** [[THIS_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store %"class.std::complex.0"* [[__C]], %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store %"class.std::complex.0"* [[__C]], %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8 // CHECK3-NEXT: [[THIS1:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: [[TMP0:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[CALL:%.*]] = call double @_ZNKSt7complexIdE4realEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP0]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL:%.*]] = call double @_ZNKSt7complexIdE4realEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP0]]) #[[ATTR9]]
// CHECK3-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[__RE_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP1:%.*]] = load double, double* [[__RE_]], align 8, !tbaa [[TBAA24:![0-9]+]] // CHECK3-NEXT: [[TMP1:%.*]] = load double, double* [[__RE_]], align 8, !tbaa [[TBAA24:![0-9]+]]
// CHECK3-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], [[CALL]] // CHECK3-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], [[CALL]]
// CHECK3-NEXT: store double [[ADD]], double* [[__RE_]], align 8, !tbaa [[TBAA24]] // CHECK3-NEXT: store double [[ADD]], double* [[__RE_]], align 8, !tbaa [[TBAA24]]
// CHECK3-NEXT: [[TMP2:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: [[TMP2:%.*]] = load %"class.std::complex.0"*, %"class.std::complex.0"** [[__C_ADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: [[CALL2:%.*]] = call double @_ZNKSt7complexIdE4imagEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]]) #[[ATTR8]] // CHECK3-NEXT: [[CALL2:%.*]] = call double @_ZNKSt7complexIdE4imagEv(%"class.std::complex.0"* nonnull align 8 dereferenceable(16) [[TMP2]]) #[[ATTR9]]
// CHECK3-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1 // CHECK3-NEXT: [[__IM_:%.*]] = getelementptr inbounds %"class.std::complex.0", %"class.std::complex.0"* [[THIS1]], i32 0, i32 1
// CHECK3-NEXT: [[TMP3:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26:![0-9]+]] // CHECK3-NEXT: [[TMP3:%.*]] = load double, double* [[__IM_]], align 8, !tbaa [[TBAA26:![0-9]+]]
// CHECK3-NEXT: [[ADD3:%.*]] = fadd double [[TMP3]], [[CALL2]] // CHECK3-NEXT: [[ADD3:%.*]] = fadd double [[TMP3]], [[CALL2]]
// CHECK3-NEXT: store double [[ADD3]], double* [[__IM_]], align 8, !tbaa [[TBAA26]] // CHECK3-NEXT: store double [[ADD3]], double* [[__IM_]], align 8, !tbaa [[TBAA26]]
// CHECK3-NEXT: ret %"class.std::complex.0"* [[THIS1]] // CHECK3-NEXT: ret %"class.std::complex.0"* [[THIS1]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func5 // CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func5
// CHECK3-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR5]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8 // CHECK3-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK3-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex.0", align 8 // CHECK3-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca %"class.std::complex.0", align 8
// CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
▲ Show 20 Lines • Show All 74 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: br label [[IFCONT6:%.*]] // CHECK3-NEXT: br label [[IFCONT6:%.*]]
// CHECK3: else5: // CHECK3: else5:
// CHECK3-NEXT: br label [[IFCONT6]] // CHECK3-NEXT: br label [[IFCONT6]]
// CHECK3: ifcont6: // CHECK3: ifcont6:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func6 // CHECK3-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func6
// CHECK3-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
// CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]] // CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8, !tbaa [[TBAA12]]
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x() // CHECK3-NEXT: [[NVPTX_TID:%.*]] = call i32 @llvm.nvvm.read.ptx.sreg.tid.x()
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK3-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK3-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4, !tbaa [[TBAA8]]
// CHECK3-NEXT: br label [[PRECOND]] // CHECK3-NEXT: br label [[PRECOND]]
// CHECK3: exit: // CHECK3: exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__3_wrapper // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__3_wrapper
// CHECK3-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i16 zeroext [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR5]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8 // CHECK3-NEXT: [[GLOBAL_ARGS:%.*]] = alloca i8**, align 8
// CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, i32* [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]] // CHECK3-NEXT: store i16 [[TMP0]], i16* [[DOTADDR]], align 2, !tbaa [[TBAA19]]
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]] // CHECK3-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4, !tbaa [[TBAA8]]
▲ Show 20 Lines • Show All 100 Lines • Show Last 20 Lines