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 SPMD Mode. 12 // CHECK-DAG: {{@__omp_offloading_.+l25}}_exec_mode = weak constant i8 0 13 // CHECK-DAG: {{@__omp_offloading_.+l30}}_exec_mode = weak constant i8 0 14 // CHECK-DAG: {{@__omp_offloading_.+l35}}_exec_mode = weak constant i8 0 15 // CHECK-DAG: {{@__omp_offloading_.+l40}}_exec_mode = weak constant i8 0 16 17 #define N 1000 18 19 template<typename tx> 20 tx ftemplate(int n) { 21 tx a[N]; 22 short aa[N]; 23 tx b[10]; 24 25 #pragma omp target simd 26 for(int i = 0; i < n; i++) { 27 a[i] = 1; 28 } 29 30 #pragma omp target simd 31 for (int i = 0; i < n; i++) { 32 aa[i] += 1; 33 } 34 35 #pragma omp target simd 36 for(int i = 0; i < 10; i++) { 37 b[i] += 1; 38 } 39 40 #pragma omp target simd reduction(+:n) 41 for(int i = 0; i < 10; i++) { 42 b[i] += 1; 43 } 44 45 return a[0]; 46 } 47 48 int bar(int n){ 49 int a = 0; 50 51 a += ftemplate<int>(n); 52 53 return a; 54 } 55 56 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l25}}( 57 // CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x() 58 // CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]], 59 // CHECK-NOT: call void @__kmpc_for_static_init 60 // CHECK-NOT: call void @__kmpc_for_static_fini 61 // CHECK: call void @__kmpc_spmd_kernel_deinit() 62 // CHECK: ret void 63 64 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l30}}( 65 // CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x() 66 // CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]], 67 // CHECK-NOT: call void @__kmpc_for_static_init 68 // CHECK-NOT: call void @__kmpc_for_static_fini 69 // CHECK: call void @__kmpc_spmd_kernel_deinit() 70 // CHECK: ret void 71 72 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l35}}( 73 // CHECK-DAG: [[THREAD_LIMIT:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.ntid.x() 74 // CHECK: call void @__kmpc_spmd_kernel_init(i32 [[THREAD_LIMIT]], 75 // CHECK-NOT: call void @__kmpc_for_static_init 76 // CHECK-NOT: call void @__kmpc_for_static_fini 77 // CHECK: call void @__kmpc_spmd_kernel_deinit() 78 // CHECK: ret void 79 80 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l40}}( 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-NOT: call void @__kmpc_for_static_init 84 // CHECK-NOT: call void @__kmpc_for_static_fini 85 // CHECK: [[RES:%.+]] = call i32 @__kmpc_nvptx_simd_reduce_nowait(i32 %{{.+}}, i32 1, i{{64|32}} {{8|4}}, i8* %{{.+}}, void (i8*, i16, i16, i16)* @{{.+}}, void (i8*, i32)* @{{.+}}) 86 // CHECK: switch i32 [[RES]] 87 // CHECK: call void @__kmpc_nvptx_end_reduce_nowait(i32 %{{.+}}) 88 // CHECK: call void @__kmpc_spmd_kernel_deinit() 89 // CHECK: ret void 90 91 92 #endif 93