clang/test/OpenMP/target_parallel_generic_loop_ast_print.cpp

   1 // RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -fopenmp \
   2 // RUN:   -fsyntax-only -verify %s
   3
   4 // RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -fopenmp \
   5 // RUN:   -ast-print %s | FileCheck %s
   6
   7 // RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -fopenmp \
   8 // RUN:   -emit-pch -o %t %s
   9
  10 // RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -fopenmp \
  11 // RUN:   -include-pch %t -ast-print %s | FileCheck %s
  12
  13 // expected-no-diagnostics
  14
  15 #ifndef HEADER
  16 #define HEADER
  17
  18 typedef void **omp_allocator_handle_t;
  19 extern const omp_allocator_handle_t omp_null_allocator;
  20 extern const omp_allocator_handle_t omp_default_mem_alloc;
  21 extern const omp_allocator_handle_t omp_large_cap_mem_alloc;
  22 extern const omp_allocator_handle_t omp_const_mem_alloc;
  23 extern const omp_allocator_handle_t omp_high_bw_mem_alloc;
  24 extern const omp_allocator_handle_t omp_low_lat_mem_alloc;
  25 extern const omp_allocator_handle_t omp_cgroup_mem_alloc;
  26 extern const omp_allocator_handle_t omp_pteam_mem_alloc;
  27 extern const omp_allocator_handle_t omp_thread_mem_alloc;
  28
  29 //CHECK: template <typename T, int C, int D> void templ_foo(T t) {
  30 //CHECK:   T j, z;
  31 //CHECK:   #pragma omp target parallel loop device(D) collapse(C) reduction(+: z) lastprivate(j) bind(thread) num_threads(C + 2)
  32 //CHECK:   for (T i = 0; i < t; ++i)
  33 //CHECK:       for (j = 0; j < t; ++j)
  34 //CHECK:           z += i + j;
  35 //CHECK: }
  36
  37 //CHECK: template<> void templ_foo<int, 2, 0>(int t) {
  38 //CHECK:     int j, z;
  39 //CHECK:     #pragma omp target parallel loop device(0) collapse(2) reduction(+: z) lastprivate(j) bind(thread) num_threads(2 + 2)
  40 //CHECK:         for (int i = 0; i < t; ++i)
  41 //CHECK:             for (j = 0; j < t; ++j)
  42 //CHECK:                 z += i + j;
  43 //CHECK: }
  44 template <typename T, int C, int D>
  45 void templ_foo(T t) {
  46
  47   T j,z;
  48   #pragma omp target parallel loop device(D) collapse(C) reduction(+:z) lastprivate(j) bind(thread) num_threads(C+2)
  49   for (T i = 0; i<t; ++i)
  50     for (j = 0; j<t; ++j)
  51       z += i+j;
  52 }
  53
  54
  55 //CHECK: void test() {
  56 void test() {
  57   constexpr int N = 100;
  58   float MTX[N][N];
  59   int aaa[1000];
  60
  61   //CHECK: #pragma omp target parallel loop map(tofrom: MTX)
  62   #pragma omp target parallel loop map(MTX)
  63   for (auto j = 0; j < N; ++j) {
  64     MTX[0][j] = 0;
  65   }
  66
  67   int j, z, z1;
  68   //CHECK: #pragma omp target parallel loop collapse(2) private(z) lastprivate(j) order(concurrent) reduction(+: z1) bind(parallel)
  69   #pragma omp target parallel loop collapse(2) private(z) lastprivate(j) \
  70                          order(concurrent) reduction(+:z1) bind(parallel)
  71   for (auto i = 0; i < N; ++i) {
  72     for (j = 0; j < N; ++j) {
  73       z = i+j;
  74       MTX[i][j] = z;
  75       z1 += z;
  76     }
  77   }
  78
  79   //CHECK: #pragma omp target parallel loop bind(thread) num_threads(16) default(none)
  80   #pragma omp target parallel loop bind(thread) num_threads(16) default(none)
  81   for (auto i = 0; i < N; ++i) { }
  82
  83   int pr;
  84   int zzz;
  85   //CHECK: #pragma omp target parallel loop private(zzz) uses_allocators(omp_default_mem_alloc) allocate(omp_default_mem_alloc: zzz) if(1) device(0) proc_bind(close) map(tofrom: pr)
  86   #pragma omp target parallel loop private(zzz) uses_allocators(omp_default_mem_alloc) allocate(omp_default_mem_alloc:zzz) if(1) device(0) proc_bind(close) map(tofrom:pr)
  87   for (int i=0; i<1000; ++i) {
  88     zzz = i + 1;
  89     pr = 33;
  90   }
  91
  92   int fpr = 10;
  93   int k;
  94   int s = 20;
  95   //CHECK: #pragma omp target parallel loop bind(thread) private(pr) firstprivate(fpr) shared(s) allocate(k) reduction(+: k)
  96   #pragma omp target parallel loop bind(thread) private(pr) firstprivate(fpr) \
  97                         shared(s) allocate(k)  reduction(+:k)
  98   for (auto i = 0; i < N; ++i) {
  99     pr = i + fpr + s;
 100   }
 101
 102   short y = 3;
 103   //CHECK: #pragma omp target parallel loop map(tofrom: y) depend(out : y)
 104   #pragma omp target parallel loop map(tofrom:y) depend(out:y)
 105   for (int i=0; i<10; ++i) {
 106     y = 3+i;
 107   }
 108 }
 109
 110 //CHECK: void nobindingfunc() {
 111 void nobindingfunc()
 112 {
 113   //CHECK: #pragma omp target parallel loop
 114   #pragma omp target parallel loop
 115   for (int i=0; i<10; ++i) { }
 116 }
 117
 118 void bar()
 119 {
 120   templ_foo<int,2,0>(8);
 121 }
 122
 123 #endif // HEADER