Files

110 строки
4.0 KiB
C++

/*
Copyright (c) 2022 Advanced Micro Devices, Inc. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
*/
// This test verifies the accuracy of hip_bfloat16 and its usage with hiprtc
#include <hip_test_common.hh>
#include <hip/hiprtc.h>
#include <hip/hip_runtime.h>
#include <hip/hip_bfloat16.h>
#include <cassert>
#include <cstddef>
#include <memory>
#include <iostream>
#include <iterator>
#include <vector>
#include <cmath>
const char* kernelname = "test_hip_bfloat16";
static constexpr auto code{
R"(
extern "C"
__global__
void test_hip_bfloat16(float* f, bool* result)
{
float &f_a = *f;
hip_bfloat16 bf_a(f_a);
float f_c = float(bf_a);
// float relative error should be less than 1/(2^7) since bfloat16
// has 7 bits mantissa.
if (fabs(f_c - f_a) / f_a <= 1.0 / 128) {
*result = true;
} else {
*result = false;
}
}
)"};
TEST_CASE("Unit_hiprtc_test_hip_bfloat16") {
using namespace std;
hiprtcProgram prog;
HIPRTC_CHECK(hiprtcCreateProgram(&prog, code, "code.cu", 0, nullptr, nullptr));
hipDeviceProp_t props;
int device = 0;
HIP_CHECK(hipGetDeviceProperties(&props, device));
#ifdef __HIP_PLATFORM_AMD__
std::string sarg = std::string("--gpu-architecture=") + props.gcnArchName;
#else
std::string sarg = std::string("--gpu-architecture=compute_") + std::to_string(props.major) +
std::to_string(props.minor);
#endif
vector<const char*> opts;
opts.push_back(sarg.c_str());
hiprtcResult compileResult{hiprtcCompileProgram(prog, opts.size(), opts.data())};
size_t logSize;
HIPRTC_CHECK(hiprtcGetProgramLogSize(prog, &logSize));
if (logSize) {
string log(logSize, '\0');
HIPRTC_CHECK(hiprtcGetProgramLog(prog, &log[0]));
std::cout << log << '\n';
}
REQUIRE(compileResult == HIPRTC_SUCCESS);
size_t codeSize;
HIPRTC_CHECK(hiprtcGetCodeSize(prog, &codeSize));
vector<char> codec(codeSize);
HIPRTC_CHECK(hiprtcGetCode(prog, codec.data()));
HIPRTC_CHECK(hiprtcDestroyProgram(&prog));
float h_a = 10.0f;
float* f_a;
bool* d_result;
HIP_CHECK(hipMalloc(&f_a, sizeof(float)));
HIP_CHECK(hipMalloc(&d_result, sizeof(bool)));
HIP_CHECK(hipMemcpy(f_a, &h_a, sizeof(float), hipMemcpyHostToDevice));
hipModule_t module;
hipFunction_t function;
HIP_CHECK(hipModuleLoadData(&module, codec.data()));
HIP_CHECK(hipModuleGetFunction(&function, module, kernelname));
struct {
float* a_;
bool* b_;
} args{f_a, d_result};
auto sizeofargs = sizeof(args);
void* config[] = {HIP_LAUNCH_PARAM_BUFFER_POINTER, &args, HIP_LAUNCH_PARAM_BUFFER_SIZE,
&sizeofargs, HIP_LAUNCH_PARAM_END};
HIP_CHECK(hipModuleLaunchKernel(function, 1, 1, 1, 1, 1, 1, 0, nullptr, nullptr, config));
HIP_CHECK(hipDeviceSynchronize());
bool h_result;
HIP_CHECK(hipMemcpyDtoH(&h_result, reinterpret_cast<hipDeviceptr_t>(d_result), sizeof(bool)));
HIP_CHECK(hipFree(d_result));
HIP_CHECK(hipModuleUnload(module));
// Result returned is true if the hip_bfloat16 accuracy is as expected
REQUIRE(h_result == true);
}