clang/test/OpenMP/target_num_teams_num_threads_attributes.cpp

   1 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm-bc %s -o %t-ppc-host.bc
   2 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple amdgcn-amd-amdhsa -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s
   3 // RUN: %clang_cc1 -target-cpu gfx900 -fopenmp -x c++ -std=c++11 -triple amdgcn-amd-amdhsa -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s
   4 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fopenmp-targets=nvptx64 -emit-llvm-bc %s -o %t-ppc-host.bc
   5 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple nvptx64 -fopenmp-targets=nvptx64 -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s
   6 // RUN: %clang_cc1 -target-cpu sm_80 -fopenmp -x c++ -std=c++11 -triple nvptx64 -fopenmp-targets=nvptx64 -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s
   7
   8 // expected-no-diagnostics
   9
  10
  11 #ifndef HEADER
  12 #define HEADER
  13
  14
  15 void default_val_num_teams() {
  16     #pragma omp target simd
  17     for (int i = 0; i < 22; i++)
  18         int a_var;
  19 }
  20
  21 void foo1() {
  22     #pragma omp target teams num_teams(22)
  23     { int a_var; }
  24 }
  25
  26 void foo2() {
  27     #pragma omp target teams distribute num_teams(22)
  28     for (int i = 0; i < 22; i++)
  29         int a_var;
  30 }
  31
  32 void foo3() {
  33     #pragma omp target teams distribute parallel for num_teams(22)
  34     for (int i = 0; i < 22; i++)
  35         int a_var;
  36 }
  37
  38 void bar1() {
  39     #pragma omp target teams num_teams(22)
  40     { int a_var; }
  41 }
  42
  43 void bar2() {
  44     #pragma omp target teams distribute num_teams(33)
  45     for (int i = 0; i < 22; i++)
  46         int a_var;
  47 }
  48
  49 void bar3() {
  50     #pragma omp target teams distribute parallel for num_teams(44)
  51     for (int i = 0; i < 22; i++)
  52         int a_var;
  53 }
  54
  55 void const_int() {
  56     const int NT = 22;
  57     #pragma omp target teams num_teams(NT)
  58     { int a_var; }
  59 }
  60
  61 void thread_limit() {
  62     #pragma omp target teams thread_limit(22)
  63     { int a_var; }
  64 }
  65
  66 void num_threads() {
  67     #pragma omp target teams distribute parallel for thread_limit(22) num_threads(11)
  68     for (int i = 0; i < 22; i++)
  69         int a_var;
  70 }
  71
  72 void threads_and_teams() {
  73     #pragma omp target teams distribute parallel for thread_limit(22) num_teams(33)
  74     for (int i = 0; i < 22; i++)
  75         int a_var;
  76 }
  77
  78 #endif
  79
  80
  81 // CHECK:      "omp_target_num_teams"="1"
  82 // CHECK:      "omp_target_num_teams"="22"
  83 // CHECK:      "omp_target_num_teams"="33"
  84 // CHECK:      "omp_target_num_teams"="44"
  85
  86 // CHECK:      "omp_target_thread_limit"="22"
  87
  88 // CHECK:      "omp_target_thread_limit"="11"
  89
  90 // CHECK:      "omp_target_num_teams"="33"
  91 // CHECK-SAME: "omp_target_thread_limit"="22"