pr881 merge changes

ddbf1db8 · Harisankar Sadasivan · f304200e · ddbf1db8 · ddbf1db8 · ddbf1db8
Commit ddbf1db8 authored Sep 15, 2023 by Harisankar Sadasivan
20 changed files
--- a/.github/CODEOWNERS
+++ b/.github/CODEOWNERS
--- a/client_example/22_grouped_gemm/CMakeLists.txt
+++ b/client_example/22_grouped_gemm/CMakeLists.txt
--- a/client_example/22_grouped_gemm/grouped_gemm_fixed_nk_fp16.cpp
+++ b/client_example/22_grouped_gemm/grouped_gemm_fixed_nk_fp16.cpp
--- a/client_example/22_grouped_gemm/grouped_gemm_fixed_nk_fp8.cpp
+++ b/client_example/22_grouped_gemm/grouped_gemm_fixed_nk_fp8.cpp
--- a/client_example/22_grouped_gemm/grouped_gemm_fixed_nk_i8.cpp
+++ b/client_example/22_grouped_gemm/grouped_gemm_fixed_nk_i8.cpp
--- a/example/15_grouped_gemm/grouped_gemm_xdl_fixed_nk_fp8.cpp
+++ b/example/15_grouped_gemm/grouped_gemm_xdl_fixed_nk_fp8.cpp
--- a/include/ck/utility/inner_product.hpp
+++ b/include/ck/utility/inner_product.hpp
@@ -10,12 +10,6 @@ namespace ck {
 template <typename TA, typename TB, typename TC>
 __device__ void inner_product(const TA& a, const TB& b, TC& c);

-template <>
-__device__ void inner_product<half_t, half_t, float>(const half_t& a, const half_t& b, float& c)
-{
-    c += a * b;
-}
-
 template <>
 __device__ void inner_product<float, float, float>(const float& a, const float& b, float& c)
 {

--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_dl_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_dl_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_gemm_fixed_nk.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_gemm_fixed_nk.hpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_km_kn_mn_irregular_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_km_kn_mn_irregular_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_km_nk_mn_irregular_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_km_nk_mn_irregular_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_dpp_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_gnwc_gkxc_gnwk_bf16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_gnwc_gkxc_gnwk_bf16_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_gnwc_gkxc_gnwk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_gnwc_gkxc_gnwk_f16_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_gnwc_gkxc_gnwk_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_gnwc_gkxc_gnwk_f32_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_nwgc_gkxc_nwgk_bf16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_nwgc_gkxc_nwgk_bf16_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_nwgc_gkxc_nwgk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_nwgc_gkxc_nwgk_f16_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_nwgc_gkxc_nwgk_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_bwd_weight/device_grouped_conv1d_bwd_weight_dl_nwgc_gkxc_nwgk_f32_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv2d_bwd_weight/device_grouped_conv2d_bwd_weight_dl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv2d_bwd_weight/device_grouped_conv2d_bwd_weight_dl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp