1 // Test target codegen - host bc file has to be created first.
2 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-cuda-mode -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc
3 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-cuda-mode -x c++ -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
4 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-cuda-mode -x c++ -triple i386-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm-bc %s -o %t-x86-host.bc
5 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-cuda-mode -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
6 // RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-cuda-mode -fexceptions -fcxx-exceptions -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
7 // expected-no-diagnostics
11 // Check that the execution mode of all 2 target regions on the gpu is set to SPMD Mode.
12 // CHECK-DAG: {{@__omp_offloading_.+l30}}_exec_mode = weak constant i8 0
13 // CHECK-DAG: {{@__omp_offloading_.+l36}}_exec_mode = weak constant i8 0
14 // CHECK-DAG: {{@__omp_offloading_.+l41}}_exec_mode = weak constant i8 0
15 // CHECK-DAG: {{@__omp_offloading_.+l46}}_exec_mode = weak constant i8 0
30 #pragma omp target teams distribute parallel for simd lastprivate(l) dist_schedule(static,128) schedule(static,32)
31 for(int i = 0; i < n; i++) {
36 #pragma omp target teams distribute parallel for simd map(tofrom: aa) num_teams(M) thread_limit(64)
37 for(int i = 0; i < n; i++) {
41 #pragma omp target teams distribute parallel for simd map(tofrom:a, aa, b) if(target: n>40) proc_bind(spread)
42 for(int i = 0; i < 10; i++) {
46 #pragma omp target teams distribute parallel for simd collapse(2) firstprivate(f) private(k)
47 for(int i = 0; i < M; i++) {
48 for(int j = 0; j < M; j++) {
60 a += ftemplate<int>(n);
65 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
66 // CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
67 // CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
68 // CHECK: call void @__kmpc_data_sharing_init_stack_spmd
69 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 91,
70 // CHECK: {{call|invoke}} void [[OUTL1:@.+]](
71 // CHECK: call void @__kmpc_for_static_fini(
72 // CHECK: call void @__kmpc_spmd_kernel_deinit()
75 // CHECK: define internal void [[OUTL1]](
76 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 33,
77 // CHECK: call void @__kmpc_for_static_fini(
80 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
81 // CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
82 // CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
83 // CHECK: call void @__kmpc_data_sharing_init_stack_spmd
84 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
85 // CHECK: {{call|invoke}} void [[OUTL2:@.+]](
86 // CHECK: call void @__kmpc_for_static_fini(
87 // CHECK: call void @__kmpc_spmd_kernel_deinit()
90 // CHECK: define internal void [[OUTL2]](
91 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 34,
92 // CHECK: call void @__kmpc_for_static_fini(
95 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}(
96 // CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
97 // CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
98 // CHECK: call void @__kmpc_data_sharing_init_stack_spmd
99 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92,
100 // CHECK: {{call|invoke}} void [[OUTL3:@.+]](
101 // CHECK: call void @__kmpc_for_static_fini(
102 // CHECK: call void @__kmpc_spmd_kernel_deinit()
105 // CHECK: define internal void [[OUTL3]](
106 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 34,
107 // CHECK: call void @__kmpc_for_static_fini(
110 // CHECK: define {{.*}}void {{@__omp_offloading_.+}}({{.+}}, i{{32|64}} [[F_IN:%.+]])
111 // CHECK: store {{.+}} [[F_IN]], {{.+}}* {{.+}},
112 // CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
113 // CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]],
114 // CHECK: call void @__kmpc_data_sharing_init_stack_spmd
115 // CHECK: store {{.+}} 99, {{.+}}* [[COMB_UB:%.+]], align
116 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92, {{.+}}, {{.+}}, {{.+}}* [[COMB_UB]],
117 // CHECK: {{call|invoke}} void [[OUTL4:@.+]](
118 // CHECK: call void @__kmpc_for_static_fini(
119 // CHECK: call void @__kmpc_spmd_kernel_deinit()
122 // CHECK: define internal void [[OUTL4]](
123 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 34,
124 // CHECK: call void @__kmpc_for_static_fini(