clang/test/Headers/gpu_disabled_math.cpp

   1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
   2 // RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h      \
   3 // RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
   4 // RUN:   -internal-isystem %S/Inputs/include \
   5 // RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
   6 // RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -o - \
   7 // RUN:   -D __CLANG_GPU_DISABLE_MATH_WRAPPERS | FileCheck -check-prefix=AMDGPU %s
   8
   9 // RUN: %clang_cc1 -include __clang_cuda_runtime_wrapper.h      \
  10 // RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
  11 // RUN:   -internal-isystem %S/Inputs/include \
  12 // RUN:   -triple nvptx64-nvidia-cuda -aux-triple x86_64-unknown-unknown \
  13 // RUN:   -target-cpu sm_90 -emit-llvm %s -fcuda-is-device -o - \
  14 // RUN:   -D __CLANG_GPU_DISABLE_MATH_WRAPPERS | FileCheck -check-prefix=NVPTX %s
  15
  16 extern "C" double sin(double x);
  17
  18 // AMDGPU-LABEL: define dso_local noundef double @_Z3food(
  19 // AMDGPU-SAME: double noundef [[X:%.*]]) #[[ATTR0:[0-9]+]] {
  20 // AMDGPU-NEXT:  [[ENTRY:.*:]]
  21 // AMDGPU-NEXT:    [[RETVAL:%.*]] = alloca double, align 8, addrspace(5)
  22 // AMDGPU-NEXT:    [[X_ADDR:%.*]] = alloca double, align 8, addrspace(5)
  23 // AMDGPU-NEXT:    [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
  24 // AMDGPU-NEXT:    [[X_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[X_ADDR]] to ptr
  25 // AMDGPU-NEXT:    store double [[X]], ptr [[X_ADDR_ASCAST]], align 8
  26 // AMDGPU-NEXT:    [[TMP0:%.*]] = load double, ptr [[X_ADDR_ASCAST]], align 8
  27 // AMDGPU-NEXT:    [[TMP1:%.*]] = call double @llvm.sin.f64(double [[TMP0]])
  28 // AMDGPU-NEXT:    ret double [[TMP1]]
  29 //
  30 // NVPTX-LABEL: define dso_local noundef double @_Z3food(
  31 // NVPTX-SAME: double noundef [[X:%.*]]) #[[ATTR0:[0-9]+]] {
  32 // NVPTX-NEXT:  [[ENTRY:.*:]]
  33 // NVPTX-NEXT:    [[X_ADDR:%.*]] = alloca double, align 8
  34 // NVPTX-NEXT:    store double [[X]], ptr [[X_ADDR]], align 8
  35 // NVPTX-NEXT:    [[TMP0:%.*]] = load double, ptr [[X_ADDR]], align 8
  36 // NVPTX-NEXT:    [[TMP1:%.*]] = call double @llvm.sin.f64(double [[TMP0]])
  37 // NVPTX-NEXT:    ret double [[TMP1]]
  38 //
  39 double foo(double x) {
  40   return sin(x);
  41 }