ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-17 11:30:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

Unified implementation of 1d/2d/3d conv bwd-data. fp32/fp16/bfp16/int8 (#134)

Browse Source 
* start convnd bwd data

* add 3d laoyout name

* add conv1d reference

* add con3d reference

* finished example client code

* conv1d kernel finished

* fix input error

* add conv3d

* add 3d layout in conv_utils.hpp

* fix sepecial check

* addconvnd lib

* add test for bwd data

* finished test

* add check slice length

* convnd bwd data start

* profiler can be compiled

* fix some bug

* set input to zero

* modify readme for example

* fix test_convnd_bwd_data bug

* test_convnd_bwd_data parameter desc

* workaround for 1d

* workaroud for 2d

* change init value

* workaround for 3d int8

* fix init value bug

* remove workaround

* fix acc data type

* add int32

* change select function to template

* tilda to tilde

* remove int32 instance

* fix commit for device hpp

* fix comments for profiler

* using profile imp to test

* add pass verification

* fix conv2d reference

* fix conflict

* remove double batched_gemm

* fix exampel conv2d data and test convnd

* format

* change conv2d_bwd_data return value

* remove repeat = 1

* remove conv bwd data

Co-authored-by: ltqin <letaoqin@amd.com>
Co-authored-by: Chao Liu <chao.liu2@amd.com>

[ROCm/composable_kernel commit: 0536f2b312]
This commit is contained in:
ltqin
2022-03-29 23:52:25 +08:00
committed by GitHub
parent 34e14a5e54
commit 7743be1e58
37 changed files with 4577 additions and 245 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
profiler/src/profiler.cpp
View File

@@ -15,7 +15,7 @@ int profile_conv_fwd(int, char*[]);
int profile_conv_fwd_bias_relu(int, char*[]);
int profile_conv_fwd_bias_relu_add(int, char*[]);
int profile_conv_fwd_bias_relu_atomic_add(int, char*[]);
int profile_conv_bwd_data(int, char*[]);
int profile_convnd_bwd_data(int, char*[], int);
int profile_reduce(int, char*[]);
int main(int argc, char* argv[])
@@ -64,9 +64,17 @@ int main(int argc, char* argv[])
{
return profile_conv_fwd_bias_relu_atomic_add(argc, argv);
}
else if(strcmp(argv[1], "conv_bwd") == 0)
else if(strcmp(argv[1], "conv1d_bwd_data") == 0)
{
return profile_conv_bwd_data(argc, argv);
return profile_convnd_bwd_data(argc, argv, 1);
}
else if(strcmp(argv[1], "conv2d_bwd_data") == 0)
{
return profile_convnd_bwd_data(argc, argv, 2);
}
else if(strcmp(argv[1], "conv3d_bwd_data") == 0)
{
return profile_convnd_bwd_data(argc, argv, 3);
}
else if(strcmp(argv[1], "reduce") == 0)
{
@@ -85,8 +93,11 @@ int main(int argc, char* argv[])
" conv_fwd_bias_relu: ForwardConvolution+Bias+ReLU\n"
" conv_fwd_bias_relu_add: ForwardConvolution+Bias+ReLU+Add\n"
" conv_fwd_bias_relu_atomic_add: ForwardConvolution+Bias+ReLU+AtomicAdd\n"
" conv_bwd: BackwardConvolution\n"
" reduce: Reduce\n");
" conv1d_bwd_data: BackwardConvolution data 1 dim\n"
" conv2d_bwd_data: BackwardConvolution data 2 dim\n"
" conv3d_bwd_data: BackwardConvolution data 3 dim\n"
" grouped_gemm: Grouped Gemm\n"
" reduce: REDUCE\n");
// clang-format on
return 0;