ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-13 09:45:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

implicit gemm v1r3 nchw_cyxk_nkhw

Browse Source
This commit is contained in:
Chao Liu
2019-04-25 15:14:39 -05:00
parent 569ad66e2a
commit a903146427
18 changed files with 1202 additions and 523 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
driver/driver.hip.cpp
View File

@@ -12,7 +12,7 @@
//#include "device_direct_convolution_2_vectorized_nchw_kcyx_nkhw.hpp"
#include "device_convolution_implicit_gemm_v1_chwn_cyxk_khwn.hpp"
#include "device_convolution_implicit_gemm_v1_nchw_cyxk_khwn.hpp"
//#include "device_implicit_gemm_convolution_1_chwn_cyxk_khwn_padded.hpp"
#include "device_convolution_implicit_gemm_v1_nchw_cyxk_nkhw.hpp"
#include "device_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hpp"
struct GeneratorTensor_1
@@ -605,8 +605,10 @@ int main(int argc, char* argv[])
device_direct_convolution_2_vectorized_nchw_kcyx_nkhw
#elif 0
device_convolution_implicit_gemm_v1_chwn_cyxk_khwn
#elif 1
#elif 0
device_convolution_implicit_gemm_v1_nchw_cyxk_khwn
#elif 1
device_convolution_implicit_gemm_v1_nchw_cyxk_nkhw
#elif 0
device_convolution_implicit_gemm_v2_chwn_cyxk_khwn
#endif