clang/test/OpenMP/amdgcn-attributes.cpp

   1 // REQUIRES: amdgpu-registered-target
   2
   3 // 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
   4 // 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 -check-prefixes=DEFAULT,ALL %s
   5 // 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 -check-prefixes=CPU,ALL %s
   6
   7 // RUN: %clang_cc1 -menable-no-nans -mno-amdgpu-ieee -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 -check-prefixes=NOIEEE,ALL %s
   8 // RUN: %clang_cc1 -munsafe-fp-atomics -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 -check-prefixes=UNSAFEATOMIC,ALL %s
   9
  10 // expected-no-diagnostics
  11
  12 #define N 100
  13
  14 int callable(int);
  15
  16 // Check that the target attributes are set on the generated kernel
  17 int func() {
  18   // ALL-LABEL: amdgpu_kernel void @__omp_offloading{{.*}} #0
  19
  20   int arr[N];
  21
  22 #pragma omp target teams thread_limit(42)
  23   for (int i = 0; i < N; i++) {
  24     arr[i] = callable(arr[i]);
  25   }
  26
  27   return arr[0];
  28 }
  29
  30 int callable(int x) {
  31   // ALL-LABEL: @_Z8callablei(i32 noundef %x) #2
  32   return x + 1;
  33 }
  34
  35 // DEFAULT: attributes #0 = { convergent mustprogress noinline norecurse nounwind optnone "amdgpu-flat-work-group-size"="1,42" "kernel" "no-trapping-math"="true" "omp_target_thread_limit"="42" "stack-protector-buffer-size"="8" "uniform-work-group-size"="true" }
  36 // CPU: attributes #0 = { convergent mustprogress noinline norecurse nounwind optnone "amdgpu-flat-work-group-size"="1,42" "kernel" "no-trapping-math"="true" "omp_target_thread_limit"="42" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" "uniform-work-group-size"="true" }
  37 // NOIEEE: attributes #0 = { convergent mustprogress noinline norecurse nounwind optnone "amdgpu-flat-work-group-size"="1,42" "amdgpu-ieee"="false" "kernel" "no-nans-fp-math"="true" "no-trapping-math"="true" "omp_target_thread_limit"="42" "stack-protector-buffer-size"="8" "uniform-work-group-size"="true" }
  38 // UNSAFEATOMIC: attributes #0 = { convergent mustprogress noinline norecurse nounwind optnone "amdgpu-flat-work-group-size"="1,42" "amdgpu-unsafe-fp-atomics"="true" "kernel" "no-trapping-math"="true" "omp_target_thread_limit"="42" "stack-protector-buffer-size"="8" "uniform-work-group-size"="true" }
  39
  40 // DEFAULT: attributes #2 = { convergent mustprogress noinline nounwind optnone "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
  41 // CPU: attributes #2 = { convergent mustprogress noinline nounwind optnone "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" }
  42 // NOIEEE: attributes #2 = { convergent mustprogress noinline nounwind optnone "amdgpu-ieee"="false" "no-nans-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
  43 // UNSAFEATOMIC: attributes #2 = { convergent mustprogress noinline nounwind optnone "amdgpu-unsafe-fp-atomics"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }