ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-14 18:17:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

Group norm (#417)

Browse Source 
* Add groupnorm example by layernorm
1.  Reference is not ready
2. shape of gamma and beta need to be fix

* Let shape of gamma and beta can be same as x

* Modify test, instance and client example

* [What] Fix bug of layernorm for greater than 2 dimension.
[Why] We need to get upper length from merge transform instead of embed transform.

* Add reference for groupnorm

* Fuse sigmoid after groupnorm

* [What] Rename original layernorm into layernorm2d
[Why] Prepare to add groupnorm using layernorm5d

* clang-format

* Add groupnorm test

* Refine error message

* Add groupnorm ckProfiler

* Test groupnorm kernel from device_instance

* update example

* upadte profiler

* Fix test naming

* Fix argc number

* Move descriptor and sweeponce to argument for quick debugging

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

[ROCm/composable_kernel commit: 4eba345f6e]
This commit is contained in:
rocking5566
2022-09-20 11:30:46 +08:00
committed by GitHub
parent 2df5929e5e
commit f09ecf09f6
24 changed files with 1218 additions and 416 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
profiler/src/profiler.cpp
View File

@@ -3,26 +3,27 @@
#include <cstring>
int profile_gemm(int, char*[]);
int profile_gemm_splitk(int, char*[]);
int profile_gemm_bilinear(int, char*[]);
int profile_gemm_add_add_fastgelu(int, char*[]);
int profile_gemm_reduce(int, char*[]);
int profile_gemm_bias_add_reduce(int, char*[]);
int profile_batched_gemm(int, char*[]);
int profile_batched_gemm_gemm(int, char*[]);
int profile_batched_gemm_add_relu_gemm_add(int, char*[]);
int profile_batched_gemm_reduce(int, char*[]);
int profile_grouped_gemm(int, char*[]);
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_bwd_data(int, char*[]);
int profile_conv_bwd_weight(int, char*[]);
int profile_grouped_conv_fwd(int, char*[]);
int profile_normalization(int, char*[]);
// int profile_gemm(int, char*[]);
// int profile_gemm_splitk(int, char*[]);
// int profile_gemm_bilinear(int, char*[]);
// int profile_gemm_add_add_fastgelu(int, char*[]);
// int profile_gemm_reduce(int, char*[]);
// int profile_gemm_bias_add_reduce(int, char*[]);
// int profile_batched_gemm(int, char*[]);
// int profile_batched_gemm_gemm(int, char*[]);
// int profile_batched_gemm_add_relu_gemm_add(int, char*[]);
// int profile_batched_gemm_reduce(int, char*[]);
// int profile_grouped_gemm(int, char*[]);
// 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_bwd_data(int, char*[]);
// int profile_conv_bwd_weight(int, char*[]);
// int profile_grouped_conv_fwd(int, char*[]);
// int profile_normalization(int, char*[]);
int profile_layernorm(int, char*[]);
int profile_reduce(int, char*[]);
int profile_groupnorm(int, char*[]);
// int profile_reduce(int, char*[]);
static void print_helper_message()
{
@@ -56,6 +57,7 @@ int main(int argc, char* argv[])
return 0;
}
#if 0
else if(strcmp(argv[1], "gemm") == 0)
{
return profile_gemm(argc, argv);
@@ -132,10 +134,15 @@ int main(int argc, char* argv[])
{
return profile_normalization(argc, argv);
}
#endif
else if(strcmp(argv[1], "layernorm") == 0)
{
return profile_layernorm(argc, argv);
}
else if(strcmp(argv[1], "groupnorm") == 0)
{
return profile_groupnorm(argc, argv);
}
else
{
print_helper_message();