ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-14 02:57:45 +00:00
Code Issues Packages Projects Releases Wiki Activity

fixed reference and host_tensor

Browse Source
This commit is contained in:
Jing Zhang
2024-10-19 19:53:17 -07:00
parent 205e0365e3
commit 05ab9105f5
4 changed files with 91 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
example/01_gemm/run_gemm_example_v2.inc
View File

@@ -158,6 +158,9 @@ bool run_gemm(const ProblemType& problem_size, const ExecutionConfig& config)
b_k_n.GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
}
b_k_n(0, 0) = 0xaa;
b_k_n(1, 1) = 0xaa;
Tensor<CDataType> c_m_n_host_result(f_host_tensor_descriptor(M, N, StrideC, CLayout{}));
Tensor<CDataType> c_m_n_device_result(f_host_tensor_descriptor(M, N, StrideC, CLayout{}));
@@ -207,31 +210,42 @@ bool run_gemm(const ProblemType& problem_size, const ExecutionConfig& config)
bool pass = true;
if(config.do_verification)
{
//auto ref_gemm = ReferenceGemmInstance{};
//auto ref_invoker = ref_gemm.MakeInvoker();
auto ref_gemm = ReferenceGemmInstance{};
auto ref_invoker = ref_gemm.MakeInvoker();
//auto ref_argument = ref_gemm.MakeArgument(
// a_m_k, b_k_n, c_m_n_host_result, PassThrough{}, PassThrough{}, PassThrough{});
auto ref_argument = ref_gemm.MakeArgument(
a_m_k, b_k_n, c_m_n_host_result, PassThrough{}, PassThrough{}, PassThrough{});
//ref_invoker.Run(ref_argument);
ref_invoker.Run(ref_argument);
ave_time = invoker.Run(argument, StreamConfig{nullptr, false, 1});
c_m_n_device_buf.FromDevice(c_m_n_device_result.mData.data());
//pass &= ck::utils::check_err(c_m_n_device_result,
// c_m_n_host_result,
// "Error: Incorrect results!",
// get_rtol<CDataType>(),
// get_atol<CDataType>());
pass &= ck::utils::check_err(c_m_n_device_result,
c_m_n_host_result,
"Error: Incorrect results!",
get_rtol<CDataType>(),
get_atol<CDataType>());
//for(int i = 0; i < M; i++)
//{
// for(int j = 0; j < N; j++)
// {
// std::cout << ck::type_convert<float>(c_m_n_device_result(i, j)) << ",";
// }
// std::cout << std::endl;
//}
std::cout << "c_m_n_device_result: " << std::endl;
for(int i = 0; i < M; i++)
{
for(int j = 0; j < N; j++)
{
std::cout << ck::type_convert<float>(c_m_n_device_result(i, j)) << ",";
}
std::cout << std::endl;
}
std::cout << "c_m_n_host_result: " << std::endl;
for(int i = 0; i < M; i++)
{
for(int j = 0; j < N; j++)
{
std::cout << ck::type_convert<float>(c_m_n_host_result(i, j)) << ",";
}
std::cout << std::endl;
}
}
if(config.time_kernel)