// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// REQUIRES: amdgpu-registered-target

// RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h      \
// RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
// RUN:   -internal-isystem %S/Inputs/include \
// RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
// RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -O1 -o - \
// RUN:   -D__HIPCC_RTC__ | FileCheck -check-prefix=DEFAULT %s

// Check that we end up with fast math flags set on intrinsic calls
// RUN: %clang_cc1 -include __clang_hip_runtime_wrapper.h \
// RUN:   -internal-isystem %S/../../lib/Headers/cuda_wrappers \
// RUN:   -internal-isystem %S/Inputs/include \
// RUN:   -triple amdgcn-amd-amdhsa -aux-triple x86_64-unknown-unknown \
// RUN:   -target-cpu gfx906 -emit-llvm %s -fcuda-is-device -O1 -ffinite-math-only -o - \
// RUN:   -D__HIPCC_RTC__ | FileCheck -check-prefix=FINITEONLY %s

// DEFAULT-LABEL: @test_fma_f16(
// DEFAULT-NEXT:  entry:
// DEFAULT-NEXT:    [[CALL_I:%.*]] = tail call contract half @__ocml_fma_f16(half noundef [[X:%.*]], half noundef [[Y:%.*]], half noundef [[Z:%.*]]) #[[ATTR4:[0-9]+]]
// DEFAULT-NEXT:    ret half [[CALL_I]]
//
// FINITEONLY-LABEL: @test_fma_f16(
// FINITEONLY-NEXT:  entry:
// FINITEONLY-NEXT:    [[CALL_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) half @__ocml_fma_f16(half noundef nofpclass(nan inf) [[X:%.*]], half noundef nofpclass(nan inf) [[Y:%.*]], half noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR4:[0-9]+]]
// FINITEONLY-NEXT:    ret half [[CALL_I]]
//
extern "C" __device__ _Float16 test_fma_f16(_Float16 x, _Float16 y,
                                            _Float16 z) {
  return fma(x, y, z);
}

// DEFAULT-LABEL: @test_pow_f16(
// DEFAULT-NEXT:  entry:
// DEFAULT-NEXT:    [[CALL_I:%.*]] = tail call contract half @__ocml_pown_f16(half noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR5:[0-9]+]]
// DEFAULT-NEXT:    ret half [[CALL_I]]
//
// FINITEONLY-LABEL: @test_pow_f16(
// FINITEONLY-NEXT:  entry:
// FINITEONLY-NEXT:    [[CALL_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) half @__ocml_pown_f16(half noundef nofpclass(nan inf) [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR5:[0-9]+]]
// FINITEONLY-NEXT:    ret half [[CALL_I]]
//
extern "C" __device__ _Float16 test_pow_f16(_Float16 x, int y) {
  return pow(x, y);
}