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 8 #ifndef HEADER 9 #define HEADER 10 11 // Check that the execution mode of all 2 target regions on the gpu is set to NonSPMD Mode. 12 // CHECK-DAG: {{@__omp_offloading_.+l30}}_exec_mode = weak constant i8 1 13 // CHECK-DAG: {{@__omp_offloading_.+l36}}_exec_mode = weak constant i8 1 14 // CHECK-DAG: {{@__omp_offloading_.+l41}}_exec_mode = weak constant i8 1 15 // CHECK-DAG: {{@__omp_offloading_.+l46}}_exec_mode = weak constant i8 1 16 17 #define N 1000 18 #define M 10 19 20 template<typename tx> 21 tx ftemplate(int n) { 22 tx a[N]; 23 short aa[N]; 24 tx b[10]; 25 tx c[M][M]; 26 tx f = n; 27 tx l; 28 int k; 29 30 #pragma omp target teams distribute simd lastprivate(l) dist_schedule(static,128) 31 for(int i = 0; i < n; i++) { 32 a[i] = 1; 33 l = i; 34 } 35 36 #pragma omp target teams distribute simd map(tofrom: aa) num_teams(M) thread_limit(64) 37 for(int i = 0; i < n; i++) { 38 aa[i] += 1; 39 } 40 41 #pragma omp target teams distribute simd map(tofrom:a, aa, b) if(target: n>40) 42 for(int i = 0; i < 10; i++) { 43 b[i] += 1; 44 } 45 46 #pragma omp target teams distribute simd collapse(2) firstprivate(f) private(k) 47 for(int i = 0; i < M; i++) { 48 for(int j = 0; j < M; j++) { 49 k = M; 50 c[i][j] = i+j*f+k; 51 } 52 } 53 54 return a[0]; 55 } 56 57 int bar(int n){ 58 int a = 0; 59 60 a += ftemplate<int>(n); 61 62 return a; 63 } 64 65 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l30( 66 // CHECK: call void @__kmpc_kernel_init(i32 %{{.+}}, i16 1) 67 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 91, 68 // CHECK: call void @__kmpc_for_static_fini( 69 // CHECK: call void @__kmpc_kernel_deinit(i16 1) 70 // CHECK: ret void 71 72 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l36( 73 // CHECK: call void @__kmpc_kernel_init(i32 %{{.+}}, i16 1) 74 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92, 75 // CHECK: call void @__kmpc_for_static_fini( 76 // CHECK: call void @__kmpc_kernel_deinit(i16 1) 77 // CHECK: ret void 78 79 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l41( 80 // CHECK: call void @__kmpc_kernel_init(i32 %{{.+}}, i16 1) 81 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92, 82 // CHECK: call void @__kmpc_for_static_fini( 83 // CHECK: call void @__kmpc_kernel_deinit(i16 1) 84 // CHECK: ret void 85 86 // CHECK: define {{.*}}void {{@__omp_offloading_.+}}_l46({{.+}}, i{{32|64}} [[F_IN:%.+]]) 87 // CHECK: store {{.+}} [[F_IN]], {{.+}}* {{.+}}, 88 // CHECK: call void @__kmpc_kernel_init(i32 %{{.+}}, i16 1) 89 // CHECK: store {{.+}} 99, {{.+}}* [[COMB_UB:%.+]], align 90 // CHECK: call void @__kmpc_for_static_init_4({{.+}}, {{.+}}, {{.+}} 92, {{.+}}, {{.+}}, {{.+}}* [[COMB_UB]], 91 // CHECK: call void @__kmpc_for_static_fini( 92 // CHECK: call void @__kmpc_kernel_deinit(i16 1) 93 // CHECK: ret void 94 95 #endif 96