46 lines · plain
1// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s2 3// CHECK: @[[EXEC_MODE1:.*]] = weak protected constant i8 14// CHECK: @llvm.compiler.used{{.*}} = appending global [1 x ptr] [ptr @[[EXEC_MODE1]]], section "llvm.metadata"5// CHECK: @[[KERNEL1_ENV:.*_kernel_environment]] = weak_odr protected constant %struct.KernelEnvironmentTy {6// CHECK-SAME: %struct.ConfigurationEnvironmentTy { i8 1, i8 1, i8 [[EXEC_MODE1:1]], i32 [[MIN_THREADS1:1]], i32 [[MAX_THREADS1:10]], i32 [[MIN_TEAMS1:1]], i32 [[MAX_TEAMS1:-1]], i32 0, i32 0 },7// CHECK-SAME: ptr @{{.*}}, ptr @{{.*}} }8 9// CHECK: @[[EXEC_MODE2:.*]] = weak protected constant i8 110// CHECK: @llvm.compiler.used{{.*}} = appending global [1 x ptr] [ptr @[[EXEC_MODE2]]], section "llvm.metadata"11// CHECK: @[[KERNEL2_ENV:.*_kernel_environment]] = weak_odr protected constant %struct.KernelEnvironmentTy {12// CHECK-SAME: %struct.ConfigurationEnvironmentTy { i8 1, i8 1, i8 [[EXEC_MODE2:1]], i32 [[MIN_THREADS2:1]], i32 [[MAX_THREADS2:30]], i32 [[MIN_TEAMS2:40]], i32 [[MAX_TEAMS2:40]], i32 0, i32 0 },13// CHECK-SAME: ptr @{{.*}}, ptr @{{.*}} }14 15module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<"dlti.alloca_memory_space", 5 : ui32>>, llvm.target_triple = "amdgcn-amd-amdhsa", omp.is_target_device = true, omp.is_gpu = true} {16 llvm.func @main(%num_teams : !llvm.ptr) {17 // CHECK: define weak_odr protected amdgpu_kernel void @__omp_offloading_{{.*}}_main_l{{[0-9]+}}(ptr %[[KERNEL_ARGS:.*]], ptr %[[NUM_TEAMS_ARG:.*]]) #[[ATTRS1:[0-9]+]]18 // CHECK: %{{.*}} = call i32 @__kmpc_target_init(ptr @[[KERNEL1_ENV]], ptr %[[KERNEL_ARGS]])19 %target_threads = llvm.mlir.constant(20) : i3220 %0 = omp.map.info var_ptr(%num_teams : !llvm.ptr, i32) map_clauses(to) capture(ByCopy) -> !llvm.ptr21 omp.target thread_limit(%target_threads : i32) map_entries(%0 -> %arg_teams : !llvm.ptr) {22 %teams_threads = llvm.mlir.constant(10) : i3223 %num_teams1 = llvm.load %arg_teams : !llvm.ptr -> i3224 omp.teams num_teams(to %num_teams1 : i32) thread_limit(%teams_threads : i32) {25 omp.terminator26 }27 omp.terminator28 }29 30 // CHECK: define weak_odr protected amdgpu_kernel void @__omp_offloading_{{.*}}_main_l{{[0-9]+}}(ptr %[[KERNEL_ARGS:.*]]) #[[ATTRS2:[0-9]+]]31 // CHECK: %{{.*}} = call i32 @__kmpc_target_init(ptr @[[KERNEL2_ENV]], ptr %[[KERNEL_ARGS]])32 %target_threads2 = llvm.mlir.constant(30) : i3233 omp.target thread_limit(%target_threads2 : i32) {34 %num_teams2 = llvm.mlir.constant(40) : i3235 omp.teams num_teams(to %num_teams2 : i32) {36 omp.terminator37 }38 omp.terminator39 }40 llvm.return41 }42}43 44// CHECK: attributes #[[ATTRS1]] = { "amdgpu-flat-work-group-size"="[[MIN_THREADS1]],[[MAX_THREADS1]]" "omp_target_thread_limit"="[[MAX_THREADS1]]" }45// CHECK: attributes #[[ATTRS2]] = { "amdgpu-flat-work-group-size"="[[MIN_THREADS2]],[[MAX_THREADS2]]" "amdgpu-max-num-workgroups"="[[MIN_TEAMS2]],1,1" "omp_target_num_teams"="[[MIN_TEAMS2]]" "omp_target_thread_limit"="[[MAX_THREADS2]]" }46