ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-13 17:55:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

Introduce ck::accumulate_n() (#439)

Browse Source 
We can use this template to eliminate duplicated iterator computing
logics. By providing return type to ck::accumulate_n(), we can avoid
type conversion operations.
This commit is contained in:
Po Yen Chen
2022-11-15 09:53:39 +08:00
committed by GitHub
parent dc663fae29
commit 730204eed0
14 changed files with 126 additions and 177 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
library/src/utility/convolution_parameter.cpp
View File

@@ -72,14 +72,10 @@ std::size_t ConvParam::GetFlops() const
{
// 2 * G * N * K * C * <output spatial lengths product> * <filter spatial lengths product>
return static_cast<std::size_t>(2) * G_ * N_ * K_ * C_ *
std::accumulate(std::begin(output_spatial_lengths_),
std::begin(output_spatial_lengths_) + num_dim_spatial_,
static_cast<std::size_t>(1),
std::multiplies<std::size_t>()) *
std::accumulate(std::begin(filter_spatial_lengths_),
std::begin(filter_spatial_lengths_) + num_dim_spatial_,
static_cast<std::size_t>(1),
std::multiplies<std::size_t>());
ck::accumulate_n<std::size_t>(
std::begin(output_spatial_lengths_), num_dim_spatial_, 1, std::multiplies<>()) *
ck::accumulate_n<std::size_t>(
std::begin(filter_spatial_lengths_), num_dim_spatial_, 1, std::multiplies<>());
}
std::string get_conv_param_parser_helper_msg()