ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-14 02:02:46 +00:00
Code Issues Packages Projects Releases Wiki Activity

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

Browse Source 
* Rangify STL algorithms

This commit adapts rangified std::copy(), std::fill() & std::transform()

* Rangify check_err()

By rangifying check_err(), we can not only compare values between
std::vector<>s, but also compare any ranges which have same value
type.

* Allow constructing Tensor<> like a HostTensorDescriptor

* Simplify Tensor<> object construction logics

* Remove more unnecessary 'HostTensorDescriptor' objects

* Re-format example code

* Re-write more HostTensorDescriptor ctor call

[ROCm/composable_kernel commit: 4a2a56c22f]
This commit is contained in:
Po Yen Chen
2022-11-12 01:36:01 +08:00
committed by GitHub
parent a4d6101e99
commit f2dd2e5b09
103 changed files with 657 additions and 649 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
example/35_splitK_gemm/run_splitK_gemm_example.inc
View File

@@ -34,15 +34,15 @@ bool run_splitK_gemm(const ProblemSize& problem_size, const ExecutionConfig& con
auto f_host_tensor_descriptor =
[](std::size_t row, std::size_t col, std::size_t stride, auto layout) {
using namespace ck::literals;
if(std::is_same<decltype(layout), ck::tensor_layout::gemm::RowMajor>::value)
{
return HostTensorDescriptor(std::vector<std::size_t>({row, col}),
std::vector<std::size_t>({stride, 1}));
return HostTensorDescriptor({row, col}, {stride, 1_uz});
}
else
{
return HostTensorDescriptor(std::vector<std::size_t>({row, col}),
std::vector<std::size_t>({1, stride}));
return HostTensorDescriptor({row, col}, {1_uz, stride});
}
};
@@ -146,15 +146,12 @@ bool run_splitK_gemm(const ProblemSize& problem_size, const ExecutionConfig& con
if(std::is_same<CDataType, ck::half_t>::value)
{
pass &= ck::utils::check_err(c_m_n_device_result.mData,
c_m_n_host_result.mData,
"fp16 incorrect result",
3e-3,
1e-3);
pass &= ck::utils::check_err(
c_m_n_device_result, c_m_n_host_result, "fp16 incorrect result", 3e-3, 1e-3);
}
else
{
pass &= ck::utils::check_err(c_m_n_device_result.mData, c_m_n_host_result.mData);
pass &= ck::utils::check_err(c_m_n_device_result, c_m_n_host_result);
}
}