caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
4794bbffb228894de3ca0dad50fc027eb94f743e
clang-p2996/clang/test/OpenMP/distribute_codegen.cpp
Joseph Huber 5300263c70 [OpenMP] Add loop tripcount argument to kernel launch and remove push function 
Previously we added the `push_target_tripcount` function to send the
loop tripcount to the device runtime so we knew how to configure the
teams / threads for execute the loop for a teams distribute construct.
This was implemented as a separate function mostly to avoid changing the
interface for backwards compatbility. Now that we've changed it anyway
and the new interface can take an arbitrary number of arguments via the
struct without changing the ABI, we can move this to the new interface.
This will simplify the runtime by removing unnecessary state between
calls.

Depends on D128550

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D128816
2022-07-08 14:44:16 -04:00

3012 lines
216 KiB
C++
Raw Blame History

// 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 _
// Test host codegen.
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-version=45 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3
// Test host codegen.
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// Test target codegen - host bc file has to be created first. (no significant differences with host version of target region)
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK17
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK17
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK19
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK19
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
void without_schedule_clause(float *a, float *b, float *c, float *d) {
#pragma omp target
#pragma omp teams
#pragma omp distribute
for (int i = 33; i < 32000000; i += 7) {
a[i] = b[i] * c[i] * d[i];
}
}
// ... loop body ...
void static_not_chunked(float *a, float *b, float *c, float *d) {
#pragma omp target
#pragma omp teams
#pragma omp distribute dist_schedule(static)
for (int i = 32000000; i > 33; i += -7) {
a[i] = b[i] * c[i] * d[i];
}
}
// ... loop body ...
void static_chunked(float *a, float *b, float *c, float *d) {
#pragma omp target
#pragma omp teams
#pragma omp distribute dist_schedule(static, 5)
for (unsigned i = 131071; i <= 2147483647; i += 127) {
a[i] = b[i] * c[i] * d[i];
}
}
// ... loop body ...
void test_precond() {
char a = 0;
#pragma omp target
#pragma omp teams
#pragma omp distribute
for(char i = a; i < 10; ++i);
}
// a is passed as a parameter to the outlined functions
// ..many loads of %0..
// no templates for now, as these require special handling in target regions and/or declare target
template <typename T>
T ftemplate() {
short aa = 0;
#pragma omp target
#pragma omp teams
#pragma omp distribute dist_schedule(static, aa)
for (int i = 0; i < 100; i++) {
}
return T();
}
int fint(void) { return ftemplate<int>(); }
#endif
// CHECK1-LABEL: define {{[^@]+}}@_Z23without_schedule_clausePfS_S_S_
// CHECK1-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK1-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK1-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK1-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load float*, float** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load float*, float** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load float*, float** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load float*, float** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to float**
// CHECK1-NEXT: store float* [[TMP0]], float** [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to float**
// CHECK1-NEXT: store float* [[TMP0]], float** [[TMP7]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP8]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to float**
// CHECK1-NEXT: store float* [[TMP1]], float** [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to float**
// CHECK1-NEXT: store float* [[TMP1]], float** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK1-NEXT: store float* [[TMP2]], float** [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float**
// CHECK1-NEXT: store float* [[TMP2]], float** [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to float**
// CHECK1-NEXT: store float* [[TMP3]], float** [[TMP20]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to float**
// CHECK1-NEXT: store float* [[TMP3]], float** [[TMP22]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP26]], align 4
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 4, i32* [[TMP27]], align 4
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP24]], i8*** [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP25]], i8*** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64** [[TMP30]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i64** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP32]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP33]], align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 4571424, i64* [[TMP34]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
// CHECK1-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56(float* [[TMP0]], float* [[TMP1]], float* [[TMP2]], float* [[TMP3]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56
// CHECK1-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK1-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK1-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK1-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined. to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[D:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store float** [[A]], float*** [[A_ADDR]], align 8
// CHECK1-NEXT: store float** [[B]], float*** [[B_ADDR]], align 8
// CHECK1-NEXT: store float** [[C]], float*** [[C_ADDR]], align 8
// CHECK1-NEXT: store float** [[D]], float*** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 33, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i64 [[IDXPROM]]
// CHECK1-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM2:%.*]] = sext i32 [[TMP16]] to i64
// CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds float, float* [[TMP15]], i64 [[IDXPROM2]]
// CHECK1-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX3]], align 4
// CHECK1-NEXT: [[MUL4:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK1-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM5:%.*]] = sext i32 [[TMP19]] to i64
// CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP18]], i64 [[IDXPROM5]]
// CHECK1-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX6]], align 4
// CHECK1-NEXT: [[MUL7:%.*]] = fmul float [[MUL4]], [[TMP20]]
// CHECK1-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP22]] to i64
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds float, float* [[TMP21]], i64 [[IDXPROM8]]
// CHECK1-NEXT: store float [[MUL7]], float* [[ARRAYIDX9]], align 4
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK1-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z18static_not_chunkedPfS_S_S_
// CHECK1-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK1-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK1-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK1-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load float*, float** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load float*, float** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load float*, float** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load float*, float** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to float**
// CHECK1-NEXT: store float* [[TMP0]], float** [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to float**
// CHECK1-NEXT: store float* [[TMP0]], float** [[TMP7]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP8]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to float**
// CHECK1-NEXT: store float* [[TMP1]], float** [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to float**
// CHECK1-NEXT: store float* [[TMP1]], float** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK1-NEXT: store float* [[TMP2]], float** [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float**
// CHECK1-NEXT: store float* [[TMP2]], float** [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to float**
// CHECK1-NEXT: store float* [[TMP3]], float** [[TMP20]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to float**
// CHECK1-NEXT: store float* [[TMP3]], float** [[TMP22]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP26]], align 4
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 4, i32* [[TMP27]], align 4
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP24]], i8*** [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP25]], i8*** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.2, i32 0, i32 0), i64** [[TMP30]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.3, i32 0, i32 0), i64** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP32]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP33]], align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 4571424, i64* [[TMP34]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
// CHECK1-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68(float* [[TMP0]], float* [[TMP1]], float* [[TMP2]], float* [[TMP3]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68
// CHECK1-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK1-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK1-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK1-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[D:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store float** [[A]], float*** [[A_ADDR]], align 8
// CHECK1-NEXT: store float** [[B]], float*** [[B_ADDR]], align 8
// CHECK1-NEXT: store float** [[C]], float*** [[C_ADDR]], align 8
// CHECK1-NEXT: store float** [[D]], float*** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 32000000, [[MUL]]
// CHECK1-NEXT: store i32 [[SUB]], i32* [[I]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i64 [[IDXPROM]]
// CHECK1-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM2:%.*]] = sext i32 [[TMP16]] to i64
// CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds float, float* [[TMP15]], i64 [[IDXPROM2]]
// CHECK1-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX3]], align 4
// CHECK1-NEXT: [[MUL4:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK1-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM5:%.*]] = sext i32 [[TMP19]] to i64
// CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP18]], i64 [[IDXPROM5]]
// CHECK1-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX6]], align 4
// CHECK1-NEXT: [[MUL7:%.*]] = fmul float [[MUL4]], [[TMP20]]
// CHECK1-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP22]] to i64
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds float, float* [[TMP21]], i64 [[IDXPROM8]]
// CHECK1-NEXT: store float [[MUL7]], float* [[ARRAYIDX9]], align 4
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z14static_chunkedPfS_S_S_
// CHECK1-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK1-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK1-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK1-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load float*, float** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load float*, float** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load float*, float** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load float*, float** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to float**
// CHECK1-NEXT: store float* [[TMP0]], float** [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to float**
// CHECK1-NEXT: store float* [[TMP0]], float** [[TMP7]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP8]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to float**
// CHECK1-NEXT: store float* [[TMP1]], float** [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to float**
// CHECK1-NEXT: store float* [[TMP1]], float** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK1-NEXT: store float* [[TMP2]], float** [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float**
// CHECK1-NEXT: store float* [[TMP2]], float** [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to float**
// CHECK1-NEXT: store float* [[TMP3]], float** [[TMP20]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to float**
// CHECK1-NEXT: store float* [[TMP3]], float** [[TMP22]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP26]], align 4
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 4, i32* [[TMP27]], align 4
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP24]], i8*** [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP25]], i8*** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64** [[TMP30]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i64** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP32]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP33]], align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 16908289, i64* [[TMP34]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
// CHECK1-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80(float* [[TMP0]], float* [[TMP1]], float* [[TMP2]], float* [[TMP3]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80
// CHECK1-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK1-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK1-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK1-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..4 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[D:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca float**, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store float** [[A]], float*** [[A_ADDR]], align 8
// CHECK1-NEXT: store float** [[B]], float*** [[B_ADDR]], align 8
// CHECK1-NEXT: store float** [[C]], float*** [[C_ADDR]], align 8
// CHECK1-NEXT: store float** [[D]], float*** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 16908288, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4u(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 5)
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK1: omp.dispatch.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp ugt i32 [[TMP6]], 16908288
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 16908288, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp ule i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK1: omp.dispatch.body:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[CMP2:%.*]] = icmp ule i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[MUL:%.*]] = mul i32 [[TMP13]], 127
// CHECK1-NEXT: [[ADD:%.*]] = add i32 131071, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[TMP14:%.*]] = load float*, float** [[TMP1]], align 8, !llvm.access.group !10
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[IDXPROM:%.*]] = zext i32 [[TMP15]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP14]], i64 [[IDXPROM]]
// CHECK1-NEXT: [[TMP16:%.*]] = load float, float* [[ARRAYIDX]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[TMP17:%.*]] = load float*, float** [[TMP2]], align 8, !llvm.access.group !10
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[IDXPROM3:%.*]] = zext i32 [[TMP18]] to i64
// CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, float* [[TMP17]], i64 [[IDXPROM3]]
// CHECK1-NEXT: [[TMP19:%.*]] = load float, float* [[ARRAYIDX4]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[MUL5:%.*]] = fmul float [[TMP16]], [[TMP19]]
// CHECK1-NEXT: [[TMP20:%.*]] = load float*, float** [[TMP3]], align 8, !llvm.access.group !10
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[IDXPROM6:%.*]] = zext i32 [[TMP21]] to i64
// CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds float, float* [[TMP20]], i64 [[IDXPROM6]]
// CHECK1-NEXT: [[TMP22:%.*]] = load float, float* [[ARRAYIDX7]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[MUL8:%.*]] = fmul float [[MUL5]], [[TMP22]]
// CHECK1-NEXT: [[TMP23:%.*]] = load float*, float** [[TMP0]], align 8, !llvm.access.group !10
// CHECK1-NEXT: [[TMP24:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[IDXPROM9:%.*]] = zext i32 [[TMP24]] to i64
// CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds float, float* [[TMP23]], i64 [[IDXPROM9]]
// CHECK1-NEXT: store float [[MUL8]], float* [[ARRAYIDX10]], align 4, !llvm.access.group !10
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !10
// CHECK1-NEXT: [[ADD11:%.*]] = add i32 [[TMP25]], 1
// CHECK1-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !10
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD12:%.*]] = add i32 [[TMP26]], [[TMP27]]
// CHECK1-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD13:%.*]] = add i32 [[TMP28]], [[TMP29]]
// CHECK1-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z12test_precondv
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK1-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i8, align 1
// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i8 0, i8* [[A]], align 1
// CHECK1-NEXT: [[TMP0:%.*]] = load i8, i8* [[A]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK1-NEXT: store i8 [[TMP0]], i8* [[CONV]], align 1
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to i64*
// CHECK1-NEXT: store i64 [[TMP1]], i64* [[TMP3]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64*
// CHECK1-NEXT: store i64 [[TMP1]], i64* [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP9:%.*]] = load i8, i8* [[A]], align 1
// CHECK1-NEXT: store i8 [[TMP9]], i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[TMP10:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP10]] to i32
// CHECK1-NEXT: [[SUB:%.*]] = sub i32 10, [[CONV2]]
// CHECK1-NEXT: [[SUB3:%.*]] = sub i32 [[SUB]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB3]], 1
// CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK1-NEXT: [[SUB4:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK1-NEXT: store i32 [[SUB4]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK1-NEXT: [[TMP12:%.*]] = zext i32 [[ADD5]] to i64
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP13]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, i32* [[TMP14]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP7]], i8*** [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP8]], i8*** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64** [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i64** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP20]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 [[TMP12]], i64* [[TMP21]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92(i64 [[TMP1]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92
// CHECK1-SAME: (i64 noundef [[A:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i8* noundef nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i8, align 1
// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i8, align 1
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I5:%.*]] = alloca i8, align 1
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK1-NEXT: store i8 [[TMP1]], i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP2]] to i32
// CHECK1-NEXT: [[SUB:%.*]] = sub i32 10, [[CONV]]
// CHECK1-NEXT: [[SUB2:%.*]] = sub i32 [[SUB]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB2]], 1
// CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK1-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK1-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: store i8 [[TMP3]], i8* [[I]], align 1
// CHECK1-NEXT: [[TMP4:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[CONV4:%.*]] = sext i8 [[TMP4]] to i32
// CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 [[CONV4]], 10
// CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK1: omp.precond.then:
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true:
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
// CHECK1-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP15:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[CONV8:%.*]] = sext i8 [[TMP15]] to i32
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[CONV8]], [[MUL]]
// CHECK1-NEXT: [[CONV10:%.*]] = trunc i32 [[ADD9]] to i8
// CHECK1-NEXT: store i8 [[CONV10]], i8* [[I5]], align 1
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK1-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]])
// CHECK1-NEXT: br label [[OMP_PRECOND_END]]
// CHECK1: omp.precond.end:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z4fintv
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z9ftemplateIiET_v()
// CHECK1-NEXT: ret i32 [[CALL]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z9ftemplateIiET_v
// CHECK1-SAME: () #[[ATTR0]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[AA_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK1-NEXT: [[TMP0:%.*]] = load i16, i16* [[AA]], align 2
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP0]], i16* [[CONV]], align 2
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to i64*
// CHECK1-NEXT: store i64 [[TMP1]], i64* [[TMP3]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64*
// CHECK1-NEXT: store i64 [[TMP1]], i64* [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP9]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, i32* [[TMP10]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP7]], i8*** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP8]], i8*** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i64** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 100, i64* [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108(i64 [[TMP1]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108
// CHECK1-SAME: (i64 noundef [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store i64 [[AA]], i64* [[AA_ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[AA_ADDR]] to i16*
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i16* [[CONV]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i16* noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i16*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i16* [[AA]], i16** [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i16*, i16** [[AA_ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i16, i16* [[TMP0]], align 2
// CHECK1-NEXT: [[CONV:%.*]] = sext i16 [[TMP1]] to i32
// CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[CONV]])
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK1: omp.dispatch.cond:
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false:
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK1: omp.dispatch.body:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !13
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !13
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK1-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !13
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK1-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK1-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK1: omp.dispatch.end:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z23without_schedule_clausePfS_S_S_
// CHECK3-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK3-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK3-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK3-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load float*, float** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load float*, float** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load float*, float** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load float*, float** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to float**
// CHECK3-NEXT: store float* [[TMP0]], float** [[TMP5]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to float**
// CHECK3-NEXT: store float* [[TMP0]], float** [[TMP7]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i8* null, i8** [[TMP8]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to float**
// CHECK3-NEXT: store float* [[TMP1]], float** [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to float**
// CHECK3-NEXT: store float* [[TMP1]], float** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK3-NEXT: store float* [[TMP2]], float** [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float**
// CHECK3-NEXT: store float* [[TMP2]], float** [[TMP17]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to float**
// CHECK3-NEXT: store float* [[TMP3]], float** [[TMP20]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to float**
// CHECK3-NEXT: store float* [[TMP3]], float** [[TMP22]], align 4
// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 4, i32* [[TMP27]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store i8** [[TMP24]], i8*** [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store i8** [[TMP25]], i8*** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64** [[TMP30]], align 4
// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i64** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store i8** null, i8*** [[TMP32]], align 4
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store i8** null, i8*** [[TMP33]], align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 4571424, i64* [[TMP34]], align 8
// CHECK3-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
// CHECK3-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56(float* [[TMP0]], float* [[TMP1]], float* [[TMP2]], float* [[TMP3]]) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56
// CHECK3-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK3-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK3-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK3-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined. to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store float** [[A]], float*** [[A_ADDR]], align 4
// CHECK3-NEXT: store float** [[B]], float*** [[B_ADDR]], align 4
// CHECK3-NEXT: store float** [[C]], float*** [[C_ADDR]], align 4
// CHECK3-NEXT: store float** [[D]], float*** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3: cond.true:
// CHECK3-NEXT: br label [[COND_END:%.*]]
// CHECK3: cond.false:
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 33, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i32 [[TMP13]]
// CHECK3-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, float* [[TMP15]], i32 [[TMP16]]
// CHECK3-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX2]], align 4
// CHECK3-NEXT: [[MUL3:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK3-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, float* [[TMP18]], i32 [[TMP19]]
// CHECK3-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX4]], align 4
// CHECK3-NEXT: [[MUL5:%.*]] = fmul float [[MUL3]], [[TMP20]]
// CHECK3-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 4
// CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP21]], i32 [[TMP22]]
// CHECK3-NEXT: store float [[MUL5]], float* [[ARRAYIDX6]], align 4
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK3-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z18static_not_chunkedPfS_S_S_
// CHECK3-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK3-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK3-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK3-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load float*, float** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load float*, float** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load float*, float** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load float*, float** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to float**
// CHECK3-NEXT: store float* [[TMP0]], float** [[TMP5]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to float**
// CHECK3-NEXT: store float* [[TMP0]], float** [[TMP7]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i8* null, i8** [[TMP8]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to float**
// CHECK3-NEXT: store float* [[TMP1]], float** [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to float**
// CHECK3-NEXT: store float* [[TMP1]], float** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK3-NEXT: store float* [[TMP2]], float** [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float**
// CHECK3-NEXT: store float* [[TMP2]], float** [[TMP17]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to float**
// CHECK3-NEXT: store float* [[TMP3]], float** [[TMP20]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to float**
// CHECK3-NEXT: store float* [[TMP3]], float** [[TMP22]], align 4
// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 4, i32* [[TMP27]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store i8** [[TMP24]], i8*** [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store i8** [[TMP25]], i8*** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.2, i32 0, i32 0), i64** [[TMP30]], align 4
// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.3, i32 0, i32 0), i64** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store i8** null, i8*** [[TMP32]], align 4
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store i8** null, i8*** [[TMP33]], align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 4571424, i64* [[TMP34]], align 8
// CHECK3-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
// CHECK3-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68(float* [[TMP0]], float* [[TMP1]], float* [[TMP2]], float* [[TMP3]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68
// CHECK3-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK3-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK3-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK3-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store float** [[A]], float*** [[A_ADDR]], align 4
// CHECK3-NEXT: store float** [[B]], float*** [[B_ADDR]], align 4
// CHECK3-NEXT: store float** [[C]], float*** [[C_ADDR]], align 4
// CHECK3-NEXT: store float** [[D]], float*** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3: cond.true:
// CHECK3-NEXT: br label [[COND_END:%.*]]
// CHECK3: cond.false:
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 32000000, [[MUL]]
// CHECK3-NEXT: store i32 [[SUB]], i32* [[I]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i32 [[TMP13]]
// CHECK3-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, float* [[TMP15]], i32 [[TMP16]]
// CHECK3-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX2]], align 4
// CHECK3-NEXT: [[MUL3:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK3-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, float* [[TMP18]], i32 [[TMP19]]
// CHECK3-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX4]], align 4
// CHECK3-NEXT: [[MUL5:%.*]] = fmul float [[MUL3]], [[TMP20]]
// CHECK3-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 4
// CHECK3-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP21]], i32 [[TMP22]]
// CHECK3-NEXT: store float [[MUL5]], float* [[ARRAYIDX6]], align 4
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z14static_chunkedPfS_S_S_
// CHECK3-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK3-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK3-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK3-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load float*, float** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load float*, float** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load float*, float** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load float*, float** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to float**
// CHECK3-NEXT: store float* [[TMP0]], float** [[TMP5]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to float**
// CHECK3-NEXT: store float* [[TMP0]], float** [[TMP7]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i8* null, i8** [[TMP8]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to float**
// CHECK3-NEXT: store float* [[TMP1]], float** [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to float**
// CHECK3-NEXT: store float* [[TMP1]], float** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK3-NEXT: store float* [[TMP2]], float** [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to float**
// CHECK3-NEXT: store float* [[TMP2]], float** [[TMP17]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to float**
// CHECK3-NEXT: store float* [[TMP3]], float** [[TMP20]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to float**
// CHECK3-NEXT: store float* [[TMP3]], float** [[TMP22]], align 4
// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 4, i32* [[TMP27]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store i8** [[TMP24]], i8*** [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store i8** [[TMP25]], i8*** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64** [[TMP30]], align 4
// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i64** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store i8** null, i8*** [[TMP32]], align 4
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store i8** null, i8*** [[TMP33]], align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 16908289, i64* [[TMP34]], align 8
// CHECK3-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
// CHECK3-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80(float* [[TMP0]], float* [[TMP1]], float* [[TMP2]], float* [[TMP3]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80
// CHECK3-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK3-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK3-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK3-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK3-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..4 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca float**, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store float** [[A]], float*** [[A_ADDR]], align 4
// CHECK3-NEXT: store float** [[B]], float*** [[B_ADDR]], align 4
// CHECK3-NEXT: store float** [[C]], float*** [[C_ADDR]], align 4
// CHECK3-NEXT: store float** [[D]], float*** [[D_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 16908288, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4u(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 5)
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK3: omp.dispatch.cond:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP:%.*]] = icmp ugt i32 [[TMP6]], 16908288
// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3: cond.true:
// CHECK3-NEXT: br label [[COND_END:%.*]]
// CHECK3: cond.false:
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 16908288, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP1:%.*]] = icmp ule i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK3: omp.dispatch.body:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[CMP2:%.*]] = icmp ule i32 [[TMP11]], [[TMP12]]
// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[MUL:%.*]] = mul i32 [[TMP13]], 127
// CHECK3-NEXT: [[ADD:%.*]] = add i32 131071, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[TMP14:%.*]] = load float*, float** [[TMP1]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP14]], i32 [[TMP15]]
// CHECK3-NEXT: [[TMP16:%.*]] = load float, float* [[ARRAYIDX]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[TMP17:%.*]] = load float*, float** [[TMP2]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds float, float* [[TMP17]], i32 [[TMP18]]
// CHECK3-NEXT: [[TMP19:%.*]] = load float, float* [[ARRAYIDX3]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[MUL4:%.*]] = fmul float [[TMP16]], [[TMP19]]
// CHECK3-NEXT: [[TMP20:%.*]] = load float*, float** [[TMP3]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds float, float* [[TMP20]], i32 [[TMP21]]
// CHECK3-NEXT: [[TMP22:%.*]] = load float, float* [[ARRAYIDX5]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[MUL6:%.*]] = fmul float [[MUL4]], [[TMP22]]
// CHECK3-NEXT: [[TMP23:%.*]] = load float*, float** [[TMP0]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[TMP24:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds float, float* [[TMP23]], i32 [[TMP24]]
// CHECK3-NEXT: store float [[MUL6]], float* [[ARRAYIDX7]], align 4, !llvm.access.group !11
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK3-NEXT: [[ADD8:%.*]] = add i32 [[TMP25]], 1
// CHECK3-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD9:%.*]] = add i32 [[TMP26]], [[TMP27]]
// CHECK3-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD10:%.*]] = add i32 [[TMP28]], [[TMP29]]
// CHECK3-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z12test_precondv
// CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK3-NEXT: [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i8, align 1
// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i8 0, i8* [[A]], align 1
// CHECK3-NEXT: [[TMP0:%.*]] = load i8, i8* [[A]], align 1
// CHECK3-NEXT: [[CONV:%.*]] = bitcast i32* [[A_CASTED]] to i8*
// CHECK3-NEXT: store i8 [[TMP0]], i8* [[CONV]], align 1
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP3]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP5]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i8* null, i8** [[TMP6]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP9:%.*]] = load i8, i8* [[A]], align 1
// CHECK3-NEXT: store i8 [[TMP9]], i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK3-NEXT: [[TMP10:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK3-NEXT: [[CONV2:%.*]] = sext i8 [[TMP10]] to i32
// CHECK3-NEXT: [[SUB:%.*]] = sub i32 10, [[CONV2]]
// CHECK3-NEXT: [[SUB3:%.*]] = sub i32 [[SUB]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add i32 [[SUB3]], 1
// CHECK3-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK3-NEXT: [[SUB4:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK3-NEXT: store i32 [[SUB4]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK3-NEXT: [[TMP12:%.*]] = zext i32 [[ADD5]] to i64
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, i32* [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, i32* [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store i8** [[TMP7]], i8*** [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store i8** [[TMP8]], i8*** [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64** [[TMP17]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i64** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store i8** null, i8*** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store i8** null, i8*** [[TMP20]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 [[TMP12]], i64* [[TMP21]], align 8
// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92(i32 [[TMP1]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92
// CHECK3-SAME: (i32 noundef [[A:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK3-NEXT: [[CONV:%.*]] = bitcast i32* [[A_ADDR]] to i8*
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i8* noundef nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i8, align 1
// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i8, align 1
// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I5:%.*]] = alloca i8, align 1
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK3-NEXT: store i8 [[TMP1]], i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK3-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK3-NEXT: [[CONV:%.*]] = sext i8 [[TMP2]] to i32
// CHECK3-NEXT: [[SUB:%.*]] = sub i32 10, [[CONV]]
// CHECK3-NEXT: [[SUB2:%.*]] = sub i32 [[SUB]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add i32 [[SUB2]], 1
// CHECK3-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK3-NEXT: store i8 [[TMP3]], i8* [[I]], align 1
// CHECK3-NEXT: [[TMP4:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK3-NEXT: [[CONV4:%.*]] = sext i8 [[TMP4]] to i32
// CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 [[CONV4]], 10
// CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK3: omp.precond.then:
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK3-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3: cond.true:
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK3-NEXT: br label [[COND_END:%.*]]
// CHECK3: cond.false:
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
// CHECK3-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP15:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK3-NEXT: [[CONV8:%.*]] = sext i8 [[TMP15]] to i32
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK3-NEXT: [[ADD9:%.*]] = add nsw i32 [[CONV8]], [[MUL]]
// CHECK3-NEXT: [[CONV10:%.*]] = trunc i32 [[ADD9]] to i8
// CHECK3-NEXT: store i8 [[CONV10]], i8* [[I5]], align 1
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK3-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]])
// CHECK3-NEXT: br label [[OMP_PRECOND_END]]
// CHECK3: omp.precond.end:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z4fintv
// CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z9ftemplateIiET_v()
// CHECK3-NEXT: ret i32 [[CALL]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z9ftemplateIiET_v
// CHECK3-SAME: () #[[ATTR0]] comdat {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[AA:%.*]] = alloca i16, align 2
// CHECK3-NEXT: [[AA_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK3-NEXT: [[TMP0:%.*]] = load i16, i16* [[AA]], align 2
// CHECK3-NEXT: [[CONV:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP0]], i16* [[CONV]], align 2
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP3]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP5]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i8* null, i8** [[TMP6]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, i32* [[TMP9]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, i32* [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store i8** [[TMP7]], i8*** [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store i8** [[TMP8]], i8*** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i64** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store i8** null, i8*** [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store i8** null, i8*** [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 100, i64* [[TMP17]], align 8
// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108(i32 [[TMP1]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108
// CHECK3-SAME: (i32 noundef [[AA:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[AA]], i32* [[AA_ADDR]], align 4
// CHECK3-NEXT: [[CONV:%.*]] = bitcast i32* [[AA_ADDR]] to i16*
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), i16* [[CONV]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i16* noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i16*, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store i16* [[AA]], i16** [[AA_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i16*, i16** [[AA_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i16, i16* [[TMP0]], align 2
// CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP1]] to i32
// CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[CONV]])
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK3: omp.dispatch.cond:
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3: cond.true:
// CHECK3-NEXT: br label [[COND_END:%.*]]
// CHECK3: cond.false:
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK3: omp.dispatch.body:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !14
// CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !14
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK3-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK3: omp.dispatch.inc:
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK3-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK3-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK3: omp.dispatch.end:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK3-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56
// CHECK17-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK17-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK17-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK17-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined. to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[D:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[C_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[D_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT: store float** [[A]], float*** [[A_ADDR]], align 8
// CHECK17-NEXT: store float** [[B]], float*** [[B_ADDR]], align 8
// CHECK17-NEXT: store float** [[C]], float*** [[C_ADDR]], align 8
// CHECK17-NEXT: store float** [[D]], float*** [[D_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 8
// CHECK17-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 8
// CHECK17-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 8
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK17: cond.true:
// CHECK17-NEXT: br label [[COND_END:%.*]]
// CHECK17: cond.false:
// CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: br label [[COND_END]]
// CHECK17: cond.end:
// CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK17: omp.inner.for.cond:
// CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK17: omp.inner.for.body:
// CHECK17-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 33, [[MUL]]
// CHECK17-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK17-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 8
// CHECK17-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i64 [[IDXPROM]]
// CHECK17-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK17-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 8
// CHECK17-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM2:%.*]] = sext i32 [[TMP16]] to i64
// CHECK17-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds float, float* [[TMP15]], i64 [[IDXPROM2]]
// CHECK17-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX3]], align 4
// CHECK17-NEXT: [[MUL4:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK17-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM5:%.*]] = sext i32 [[TMP19]] to i64
// CHECK17-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP18]], i64 [[IDXPROM5]]
// CHECK17-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX6]], align 4
// CHECK17-NEXT: [[MUL7:%.*]] = fmul float [[MUL4]], [[TMP20]]
// CHECK17-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 8
// CHECK17-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP22]] to i64
// CHECK17-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds float, float* [[TMP21]], i64 [[IDXPROM8]]
// CHECK17-NEXT: store float [[MUL7]], float* [[ARRAYIDX9]], align 4
// CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK17: omp.body.continue:
// CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK17: omp.inner.for.inc:
// CHECK17-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK17-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK17: omp.inner.for.end:
// CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK17: omp.loop.exit:
// CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68
// CHECK17-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK17-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK17-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK17-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[D:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[C_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[D_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT: store float** [[A]], float*** [[A_ADDR]], align 8
// CHECK17-NEXT: store float** [[B]], float*** [[B_ADDR]], align 8
// CHECK17-NEXT: store float** [[C]], float*** [[C_ADDR]], align 8
// CHECK17-NEXT: store float** [[D]], float*** [[D_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 8
// CHECK17-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 8
// CHECK17-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 8
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK17: cond.true:
// CHECK17-NEXT: br label [[COND_END:%.*]]
// CHECK17: cond.false:
// CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: br label [[COND_END]]
// CHECK17: cond.end:
// CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK17: omp.inner.for.cond:
// CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK17: omp.inner.for.body:
// CHECK17-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK17-NEXT: [[SUB:%.*]] = sub nsw i32 32000000, [[MUL]]
// CHECK17-NEXT: store i32 [[SUB]], i32* [[I]], align 4
// CHECK17-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 8
// CHECK17-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i64 [[IDXPROM]]
// CHECK17-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK17-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 8
// CHECK17-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM2:%.*]] = sext i32 [[TMP16]] to i64
// CHECK17-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds float, float* [[TMP15]], i64 [[IDXPROM2]]
// CHECK17-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX3]], align 4
// CHECK17-NEXT: [[MUL4:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK17-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM5:%.*]] = sext i32 [[TMP19]] to i64
// CHECK17-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP18]], i64 [[IDXPROM5]]
// CHECK17-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX6]], align 4
// CHECK17-NEXT: [[MUL7:%.*]] = fmul float [[MUL4]], [[TMP20]]
// CHECK17-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 8
// CHECK17-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK17-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP22]] to i64
// CHECK17-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds float, float* [[TMP21]], i64 [[IDXPROM8]]
// CHECK17-NEXT: store float [[MUL7]], float* [[ARRAYIDX9]], align 4
// CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK17: omp.body.continue:
// CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK17: omp.inner.for.inc:
// CHECK17-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK17-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK17: omp.inner.for.end:
// CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK17: omp.loop.exit:
// CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80
// CHECK17-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[C_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[D_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK17-NEXT: store float* [[B]], float** [[B_ADDR]], align 8
// CHECK17-NEXT: store float* [[C]], float** [[C_ADDR]], align 8
// CHECK17-NEXT: store float* [[D]], float** [[D_ADDR]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..2 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[A:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[B:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[C:%.*]], float** noundef nonnull align 8 dereferenceable(8) [[D:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[C_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[D_ADDR:%.*]] = alloca float**, align 8
// CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT: store float** [[A]], float*** [[A_ADDR]], align 8
// CHECK17-NEXT: store float** [[B]], float*** [[B_ADDR]], align 8
// CHECK17-NEXT: store float** [[C]], float*** [[C_ADDR]], align 8
// CHECK17-NEXT: store float** [[D]], float*** [[D_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 8
// CHECK17-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 8
// CHECK17-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 8
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 16908288, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK17-NEXT: call void @__kmpc_for_static_init_4u(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 5)
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK17: omp.dispatch.cond:
// CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP:%.*]] = icmp ugt i32 [[TMP6]], 16908288
// CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK17: cond.true:
// CHECK17-NEXT: br label [[COND_END:%.*]]
// CHECK17: cond.false:
// CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: br label [[COND_END]]
// CHECK17: cond.end:
// CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 16908288, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP1:%.*]] = icmp ule i32 [[TMP9]], [[TMP10]]
// CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK17: omp.dispatch.body:
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK17: omp.inner.for.cond:
// CHECK17-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[CMP2:%.*]] = icmp ule i32 [[TMP11]], [[TMP12]]
// CHECK17-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK17: omp.inner.for.body:
// CHECK17-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[MUL:%.*]] = mul i32 [[TMP13]], 127
// CHECK17-NEXT: [[ADD:%.*]] = add i32 131071, [[MUL]]
// CHECK17-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[TMP14:%.*]] = load float*, float** [[TMP1]], align 8, !llvm.access.group !11
// CHECK17-NEXT: [[TMP15:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[IDXPROM:%.*]] = zext i32 [[TMP15]] to i64
// CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP14]], i64 [[IDXPROM]]
// CHECK17-NEXT: [[TMP16:%.*]] = load float, float* [[ARRAYIDX]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[TMP17:%.*]] = load float*, float** [[TMP2]], align 8, !llvm.access.group !11
// CHECK17-NEXT: [[TMP18:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[IDXPROM3:%.*]] = zext i32 [[TMP18]] to i64
// CHECK17-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, float* [[TMP17]], i64 [[IDXPROM3]]
// CHECK17-NEXT: [[TMP19:%.*]] = load float, float* [[ARRAYIDX4]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[MUL5:%.*]] = fmul float [[TMP16]], [[TMP19]]
// CHECK17-NEXT: [[TMP20:%.*]] = load float*, float** [[TMP3]], align 8, !llvm.access.group !11
// CHECK17-NEXT: [[TMP21:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[IDXPROM6:%.*]] = zext i32 [[TMP21]] to i64
// CHECK17-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds float, float* [[TMP20]], i64 [[IDXPROM6]]
// CHECK17-NEXT: [[TMP22:%.*]] = load float, float* [[ARRAYIDX7]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[MUL8:%.*]] = fmul float [[MUL5]], [[TMP22]]
// CHECK17-NEXT: [[TMP23:%.*]] = load float*, float** [[TMP0]], align 8, !llvm.access.group !11
// CHECK17-NEXT: [[TMP24:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[IDXPROM9:%.*]] = zext i32 [[TMP24]] to i64
// CHECK17-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds float, float* [[TMP23]], i64 [[IDXPROM9]]
// CHECK17-NEXT: store float [[MUL8]], float* [[ARRAYIDX10]], align 4, !llvm.access.group !11
// CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK17: omp.body.continue:
// CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK17: omp.inner.for.inc:
// CHECK17-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK17-NEXT: [[ADD11:%.*]] = add i32 [[TMP25]], 1
// CHECK17-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK17: omp.inner.for.end:
// CHECK17-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: [[ADD12:%.*]] = add i32 [[TMP26]], [[TMP27]]
// CHECK17-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: [[ADD13:%.*]] = add i32 [[TMP28]], [[TMP29]]
// CHECK17-NEXT: store i32 [[ADD13]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
// CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92
// CHECK17-SAME: (i64 noundef [[A:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i8* noundef nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[TMP:%.*]] = alloca i8, align 1
// CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK17-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[I:%.*]] = alloca i8, align 1
// CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[I5:%.*]] = alloca i8, align 1
// CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK17-NEXT: [[TMP1:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK17-NEXT: store i8 [[TMP1]], i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK17-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK17-NEXT: [[CONV:%.*]] = sext i8 [[TMP2]] to i32
// CHECK17-NEXT: [[SUB:%.*]] = sub i32 10, [[CONV]]
// CHECK17-NEXT: [[SUB2:%.*]] = sub i32 [[SUB]], 1
// CHECK17-NEXT: [[ADD:%.*]] = add i32 [[SUB2]], 1
// CHECK17-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK17-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK17-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK17-NEXT: [[TMP3:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK17-NEXT: store i8 [[TMP3]], i8* [[I]], align 1
// CHECK17-NEXT: [[TMP4:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK17-NEXT: [[CONV4:%.*]] = sext i8 [[TMP4]] to i32
// CHECK17-NEXT: [[CMP:%.*]] = icmp slt i32 [[CONV4]], 10
// CHECK17-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK17: omp.precond.then:
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK17-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK17-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
// CHECK17-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK17: cond.true:
// CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK17-NEXT: br label [[COND_END:%.*]]
// CHECK17: cond.false:
// CHECK17-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: br label [[COND_END]]
// CHECK17: cond.end:
// CHECK17-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
// CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK17: omp.inner.for.cond:
// CHECK17-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
// CHECK17-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK17: omp.inner.for.body:
// CHECK17-NEXT: [[TMP15:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK17-NEXT: [[CONV8:%.*]] = sext i8 [[TMP15]] to i32
// CHECK17-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK17-NEXT: [[ADD9:%.*]] = add nsw i32 [[CONV8]], [[MUL]]
// CHECK17-NEXT: [[CONV10:%.*]] = trunc i32 [[ADD9]] to i8
// CHECK17-NEXT: store i8 [[CONV10]], i8* [[I5]], align 1
// CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK17: omp.body.continue:
// CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK17: omp.inner.for.inc:
// CHECK17-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK17-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK17: omp.inner.for.end:
// CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK17: omp.loop.exit:
// CHECK17-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4
// CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]])
// CHECK17-NEXT: br label [[OMP_PRECOND_END]]
// CHECK17: omp.precond.end:
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108
// CHECK17-SAME: (i64 noundef [[AA:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: store i64 [[AA]], i64* [[AA_ADDR]], align 8
// CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[AA_ADDR]] to i16*
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i16* [[CONV]])
// CHECK17-NEXT: ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i16* noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i16*, align 8
// CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT: store i16* [[AA]], i16** [[AA_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load i16*, i16** [[AA_ADDR]], align 8
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK17-NEXT: [[TMP1:%.*]] = load i16, i16* [[TMP0]], align 2
// CHECK17-NEXT: [[CONV:%.*]] = sext i16 [[TMP1]] to i32
// CHECK17-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[CONV]])
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK17: omp.dispatch.cond:
// CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK17: cond.true:
// CHECK17-NEXT: br label [[COND_END:%.*]]
// CHECK17: cond.false:
// CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: br label [[COND_END]]
// CHECK17: cond.end:
// CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK17: omp.dispatch.body:
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK17: omp.inner.for.cond:
// CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !14
// CHECK17-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK17-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK17: omp.inner.for.body:
// CHECK17-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
// CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK17-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !14
// CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK17: omp.body.continue:
// CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK17: omp.inner.for.inc:
// CHECK17-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK17-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK17-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK17: omp.inner.for.end:
// CHECK17-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK17: omp.dispatch.inc:
// CHECK17-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK17-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_LB]], align 4
// CHECK17-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK17-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK17-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_UB]], align 4
// CHECK17-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK17: omp.dispatch.end:
// CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK17-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23without_schedule_clausePfS_S_S__l56
// CHECK19-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK19-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK19-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK19-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined. to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[C_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[D_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK19-NEXT: store float** [[A]], float*** [[A_ADDR]], align 4
// CHECK19-NEXT: store float** [[B]], float*** [[B_ADDR]], align 4
// CHECK19-NEXT: store float** [[C]], float*** [[C_ADDR]], align 4
// CHECK19-NEXT: store float** [[D]], float*** [[D_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK19: cond.true:
// CHECK19-NEXT: br label [[COND_END:%.*]]
// CHECK19: cond.false:
// CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: br label [[COND_END]]
// CHECK19: cond.end:
// CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK19: omp.inner.for.cond:
// CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK19: omp.inner.for.body:
// CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 33, [[MUL]]
// CHECK19-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK19-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 4
// CHECK19-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i32 [[TMP13]]
// CHECK19-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 4
// CHECK19-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, float* [[TMP15]], i32 [[TMP16]]
// CHECK19-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX2]], align 4
// CHECK19-NEXT: [[MUL3:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK19-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, float* [[TMP18]], i32 [[TMP19]]
// CHECK19-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX4]], align 4
// CHECK19-NEXT: [[MUL5:%.*]] = fmul float [[MUL3]], [[TMP20]]
// CHECK19-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 4
// CHECK19-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP21]], i32 [[TMP22]]
// CHECK19-NEXT: store float [[MUL5]], float* [[ARRAYIDX6]], align 4
// CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK19: omp.body.continue:
// CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK19: omp.inner.for.inc:
// CHECK19-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK19-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK19: omp.inner.for.end:
// CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK19: omp.loop.exit:
// CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18static_not_chunkedPfS_S_S__l68
// CHECK19-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK19-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK19-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK19-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..1 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[C_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[D_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK19-NEXT: store float** [[A]], float*** [[A_ADDR]], align 4
// CHECK19-NEXT: store float** [[B]], float*** [[B_ADDR]], align 4
// CHECK19-NEXT: store float** [[C]], float*** [[C_ADDR]], align 4
// CHECK19-NEXT: store float** [[D]], float*** [[D_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 4571423, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 4571423
// CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK19: cond.true:
// CHECK19-NEXT: br label [[COND_END:%.*]]
// CHECK19: cond.false:
// CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: br label [[COND_END]]
// CHECK19: cond.end:
// CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 4571423, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK19: omp.inner.for.cond:
// CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK19: omp.inner.for.body:
// CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 7
// CHECK19-NEXT: [[SUB:%.*]] = sub nsw i32 32000000, [[MUL]]
// CHECK19-NEXT: store i32 [[SUB]], i32* [[I]], align 4
// CHECK19-NEXT: [[TMP12:%.*]] = load float*, float** [[TMP1]], align 4
// CHECK19-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP12]], i32 [[TMP13]]
// CHECK19-NEXT: [[TMP14:%.*]] = load float, float* [[ARRAYIDX]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = load float*, float** [[TMP2]], align 4
// CHECK19-NEXT: [[TMP16:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, float* [[TMP15]], i32 [[TMP16]]
// CHECK19-NEXT: [[TMP17:%.*]] = load float, float* [[ARRAYIDX2]], align 4
// CHECK19-NEXT: [[MUL3:%.*]] = fmul float [[TMP14]], [[TMP17]]
// CHECK19-NEXT: [[TMP18:%.*]] = load float*, float** [[TMP3]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, float* [[TMP18]], i32 [[TMP19]]
// CHECK19-NEXT: [[TMP20:%.*]] = load float, float* [[ARRAYIDX4]], align 4
// CHECK19-NEXT: [[MUL5:%.*]] = fmul float [[MUL3]], [[TMP20]]
// CHECK19-NEXT: [[TMP21:%.*]] = load float*, float** [[TMP0]], align 4
// CHECK19-NEXT: [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK19-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, float* [[TMP21]], i32 [[TMP22]]
// CHECK19-NEXT: store float [[MUL5]], float* [[ARRAYIDX6]], align 4
// CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK19: omp.body.continue:
// CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK19: omp.inner.for.inc:
// CHECK19-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK19-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK19: omp.inner.for.end:
// CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK19: omp.loop.exit:
// CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14static_chunkedPfS_S_S__l80
// CHECK19-SAME: (float* noundef [[A:%.*]], float* noundef [[B:%.*]], float* noundef [[C:%.*]], float* noundef [[D:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[C_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[D_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK19-NEXT: store float* [[B]], float** [[B_ADDR]], align 4
// CHECK19-NEXT: store float* [[C]], float** [[C_ADDR]], align 4
// CHECK19-NEXT: store float* [[D]], float** [[D_ADDR]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* @.omp_outlined..2 to void (i32*, i32*, ...)*), float** [[A_ADDR]], float** [[B_ADDR]], float** [[C_ADDR]], float** [[D_ADDR]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[A:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[B:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[C:%.*]], float** noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[C_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[D_ADDR:%.*]] = alloca float**, align 4
// CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK19-NEXT: store float** [[A]], float*** [[A_ADDR]], align 4
// CHECK19-NEXT: store float** [[B]], float*** [[B_ADDR]], align 4
// CHECK19-NEXT: store float** [[C]], float*** [[C_ADDR]], align 4
// CHECK19-NEXT: store float** [[D]], float*** [[D_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load float**, float*** [[A_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load float**, float*** [[B_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load float**, float*** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load float**, float*** [[D_ADDR]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 16908288, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK19-NEXT: call void @__kmpc_for_static_init_4u(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 5)
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK19: omp.dispatch.cond:
// CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP:%.*]] = icmp ugt i32 [[TMP6]], 16908288
// CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK19: cond.true:
// CHECK19-NEXT: br label [[COND_END:%.*]]
// CHECK19: cond.false:
// CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: br label [[COND_END]]
// CHECK19: cond.end:
// CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 16908288, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP1:%.*]] = icmp ule i32 [[TMP9]], [[TMP10]]
// CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK19: omp.dispatch.body:
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK19: omp.inner.for.cond:
// CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[CMP2:%.*]] = icmp ule i32 [[TMP11]], [[TMP12]]
// CHECK19-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK19: omp.inner.for.body:
// CHECK19-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[MUL:%.*]] = mul i32 [[TMP13]], 127
// CHECK19-NEXT: [[ADD:%.*]] = add i32 131071, [[MUL]]
// CHECK19-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[TMP14:%.*]] = load float*, float** [[TMP1]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[TMP15:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, float* [[TMP14]], i32 [[TMP15]]
// CHECK19-NEXT: [[TMP16:%.*]] = load float, float* [[ARRAYIDX]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[TMP17:%.*]] = load float*, float** [[TMP2]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[TMP18:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds float, float* [[TMP17]], i32 [[TMP18]]
// CHECK19-NEXT: [[TMP19:%.*]] = load float, float* [[ARRAYIDX3]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[MUL4:%.*]] = fmul float [[TMP16]], [[TMP19]]
// CHECK19-NEXT: [[TMP20:%.*]] = load float*, float** [[TMP3]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[TMP21:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds float, float* [[TMP20]], i32 [[TMP21]]
// CHECK19-NEXT: [[TMP22:%.*]] = load float, float* [[ARRAYIDX5]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[MUL6:%.*]] = fmul float [[MUL4]], [[TMP22]]
// CHECK19-NEXT: [[TMP23:%.*]] = load float*, float** [[TMP0]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[TMP24:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds float, float* [[TMP23]], i32 [[TMP24]]
// CHECK19-NEXT: store float [[MUL6]], float* [[ARRAYIDX7]], align 4, !llvm.access.group !12
// CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK19: omp.body.continue:
// CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK19: omp.inner.for.inc:
// CHECK19-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK19-NEXT: [[ADD8:%.*]] = add i32 [[TMP25]], 1
// CHECK19-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK19: omp.inner.for.end:
// CHECK19-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: [[ADD9:%.*]] = add i32 [[TMP26]], [[TMP27]]
// CHECK19-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: [[ADD10:%.*]] = add i32 [[TMP28]], [[TMP29]]
// CHECK19-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
// CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z12test_precondv_l92
// CHECK19-SAME: (i32 noundef [[A:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK19-NEXT: [[CONV:%.*]] = bitcast i32* [[A_ADDR]] to i8*
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i8* noundef nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 4
// CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[TMP:%.*]] = alloca i8, align 1
// CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK19-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[I:%.*]] = alloca i8, align 1
// CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[I5:%.*]] = alloca i8, align 1
// CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK19-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK19-NEXT: store i8 [[TMP1]], i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK19-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK19-NEXT: [[CONV:%.*]] = sext i8 [[TMP2]] to i32
// CHECK19-NEXT: [[SUB:%.*]] = sub i32 10, [[CONV]]
// CHECK19-NEXT: [[SUB2:%.*]] = sub i32 [[SUB]], 1
// CHECK19-NEXT: [[ADD:%.*]] = add i32 [[SUB2]], 1
// CHECK19-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK19-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK19-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK19-NEXT: store i8 [[TMP3]], i8* [[I]], align 1
// CHECK19-NEXT: [[TMP4:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK19-NEXT: [[CONV4:%.*]] = sext i8 [[TMP4]] to i32
// CHECK19-NEXT: [[CMP:%.*]] = icmp slt i32 [[CONV4]], 10
// CHECK19-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK19: omp.precond.then:
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK19-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK19-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
// CHECK19-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK19: cond.true:
// CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK19-NEXT: br label [[COND_END:%.*]]
// CHECK19: cond.false:
// CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: br label [[COND_END]]
// CHECK19: cond.end:
// CHECK19-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
// CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK19: omp.inner.for.cond:
// CHECK19-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
// CHECK19-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK19: omp.inner.for.body:
// CHECK19-NEXT: [[TMP15:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK19-NEXT: [[CONV8:%.*]] = sext i8 [[TMP15]] to i32
// CHECK19-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK19-NEXT: [[ADD9:%.*]] = add nsw i32 [[CONV8]], [[MUL]]
// CHECK19-NEXT: [[CONV10:%.*]] = trunc i32 [[ADD9]] to i8
// CHECK19-NEXT: store i8 [[CONV10]], i8* [[I5]], align 1
// CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK19: omp.body.continue:
// CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK19: omp.inner.for.inc:
// CHECK19-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK19-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK19: omp.inner.for.end:
// CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK19: omp.loop.exit:
// CHECK19-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4
// CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]])
// CHECK19-NEXT: br label [[OMP_PRECOND_END]]
// CHECK19: omp.precond.end:
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_v_l108
// CHECK19-SAME: (i32 noundef [[AA:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store i32 [[AA]], i32* [[AA_ADDR]], align 4
// CHECK19-NEXT: [[CONV:%.*]] = bitcast i32* [[AA_ADDR]] to i16*
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i16*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i16* [[CONV]])
// CHECK19-NEXT: ret void
//
//
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i16* noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i16*, align 4
// CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK19-NEXT: store i16* [[AA]], i16** [[AA_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load i16*, i16** [[AA_ADDR]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i16, i16* [[TMP0]], align 2
// CHECK19-NEXT: [[CONV:%.*]] = sext i16 [[TMP1]] to i32
// CHECK19-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]], i32 91, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 [[CONV]])
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND:%.*]]
// CHECK19: omp.dispatch.cond:
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK19: cond.true:
// CHECK19-NEXT: br label [[COND_END:%.*]]
// CHECK19: cond.false:
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: br label [[COND_END]]
// CHECK19: cond.end:
// CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]]
// CHECK19: omp.dispatch.body:
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK19: omp.inner.for.cond:
// CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !15
// CHECK19-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK19-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK19: omp.inner.for.body:
// CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
// CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK19-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !15
// CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK19: omp.body.continue:
// CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK19: omp.inner.for.inc:
// CHECK19-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK19-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK19: omp.inner.for.end:
// CHECK19-NEXT: br label [[OMP_DISPATCH_INC:%.*]]
// CHECK19: omp.dispatch.inc:
// CHECK19-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK19-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_LB]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK19-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK19-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_UB]], align 4
// CHECK19-NEXT: br label [[OMP_DISPATCH_COND]]
// CHECK19: omp.dispatch.end:
// CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK19-NEXT: ret void
//
Reference in New Issue View Git Blame Copy Permalink