Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
gaoqiong
composable_kernel
Commits
adfcb029
"example/13_conv2d_backward_weight_xdl/main.cpp" did not exist on "6dfb92bbef33b4caea55f6b4ed7c449927ae771c"
Commit
adfcb029
authored
Sep 22, 2023
by
Jing Zhang
Browse files
fixed merge
parents
141efec8
bba085d2
Changes
99
Hide whitespace changes
Inline
Side-by-side
Showing
20 changed files
with
167 additions
and
262 deletions
+167
-262
library/src/tensor_operation_instance/gpu/contraction_bilinear/CMakeLists.txt
...peration_instance/gpu/contraction_bilinear/CMakeLists.txt
+12
-15
library/src/tensor_operation_instance/gpu/contraction_scale/CMakeLists.txt
...r_operation_instance/gpu/contraction_scale/CMakeLists.txt
+12
-14
library/src/tensor_operation_instance/gpu/conv2d_bwd_data/CMakeLists.txt
...sor_operation_instance/gpu/conv2d_bwd_data/CMakeLists.txt
+8
-21
library/src/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt
...c/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt
+6
-14
library/src/tensor_operation_instance/gpu/elementwise_normalization/CMakeLists.txt
...ion_instance/gpu/elementwise_normalization/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt
...ary/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt
+93
-107
library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/CMakeLists.txt
...eration_instance/gpu/gemm_add_add_fastgelu/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/gemm_add_fastgelu/CMakeLists.txt
...r_operation_instance/gpu/gemm_add_fastgelu/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/gemm_add_relu_add_layernorm/CMakeLists.txt
...n_instance/gpu/gemm_add_relu_add_layernorm/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/gemm_bilinear/CMakeLists.txt
...ensor_operation_instance/gpu/gemm_bilinear/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/gemm_fastgelu/CMakeLists.txt
...ensor_operation_instance/gpu/gemm_fastgelu/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/gemm_multiply_add/CMakeLists.txt
...r_operation_instance/gpu/gemm_multiply_add/CMakeLists.txt
+4
-11
library/src/tensor_operation_instance/gpu/gemm_splitk/CMakeLists.txt
.../tensor_operation_instance/gpu/gemm_splitk/CMakeLists.txt
+16
-24
library/src/tensor_operation_instance/gpu/gemm_streamk/CMakeLists.txt
...tensor_operation_instance/gpu/gemm_streamk/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
..._operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
+1
-10
library/src/tensor_operation_instance/gpu/grouped_gemm/CMakeLists.txt
...tensor_operation_instance/gpu/grouped_gemm/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/grouped_gemm_fastgelu/CMakeLists.txt
...eration_instance/gpu/grouped_gemm_fastgelu/CMakeLists.txt
+0
-2
library/src/tensor_operation_instance/gpu/max_pool_bwd/CMakeLists.txt
...tensor_operation_instance/gpu/max_pool_bwd/CMakeLists.txt
+3
-9
library/src/tensor_operation_instance/gpu/normalization/CMakeLists.txt
...ensor_operation_instance/gpu/normalization/CMakeLists.txt
+6
-7
library/src/tensor_operation_instance/gpu/pool3d_fwd/CMakeLists.txt
...c/tensor_operation_instance/gpu/pool3d_fwd/CMakeLists.txt
+6
-12
No files found.
library/src/tensor_operation_instance/gpu/contraction_bilinear/CMakeLists.txt
View file @
adfcb029
set
(
DEVICE_CONTRACTION_BILINEAR_INSTANCES
)
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
#float
list
(
APPEND DEVICE_CONTRACTION_BILINEAR_INSTANCES device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp64"
OR NOT DEFINED DTYPES
)
#double
list
(
APPEND DEVICE_CONTRACTION_BILINEAR_INSTANCES device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_kknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_knnn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_mknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_mnnn_instance.cpp
)
endif
()
add_instance_library
(
device_contraction_bilinear_instance
${
DEVICE_CONTRACTION_BILINEAR_INSTANCES
}
)
#float
list
(
APPEND DEVICE_CONTRACTION_BILINEAR_INSTANCES device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn_instance.cpp
)
#double
list
(
APPEND DEVICE_CONTRACTION_BILINEAR_INSTANCES device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_kknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_knnn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_mknn_instance.cpp
device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_f64_mnnn_instance.cpp
)
add_instance_library
(
device_contraction_bilinear_instance
${
DEVICE_CONTRACTION_BILINEAR_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/contraction_scale/CMakeLists.txt
View file @
adfcb029
set
(
DEVICE_CONTRACTION_SCALE_INSTANCES
)
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
#float
list
(
APPEND DEVICE_CONTRACTION_SCALE_INSTANCES device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp64"
OR NOT DEFINED DTYPES
)
#double
list
(
APPEND DEVICE_CONTRACTION_SCALE_INSTANCES device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_kkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_knn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_mkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_mnn_instance.cpp
)
endif
()
#float
list
(
APPEND DEVICE_CONTRACTION_SCALE_INSTANCES device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instance.cpp
)
#double
list
(
APPEND DEVICE_CONTRACTION_SCALE_INSTANCES device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_kkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_knn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_mkn_instance.cpp
device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f64_f64_f64_mnn_instance.cpp
)
add_instance_library
(
device_contraction_scale_instance
${
DEVICE_CONTRACTION_SCALE_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/conv2d_bwd_data/CMakeLists.txt
View file @
adfcb029
set
(
CONV2D_BWD_DATA_INSTANCES
)
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
)
if
(
DL_KERNELS
)
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f32_instance.cpp
)
endif
()
endif
()
if
(
DTYPES MATCHES
"bf16"
OR NOT DEFINED DTYPES
)
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
)
if
(
DL_KERNELS
)
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f16_instance.cpp
)
endif
()
endif
()
if
(
DTYPES MATCHES
"int8"
OR NOT DEFINED DTYPES
)
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
)
if
(
DL_KERNELS
)
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_int8_instance.cpp
)
endif
()
endif
()
list
(
APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f32_instance.cpp
device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f16_instance.cpp
device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_int8_instance.cpp
)
add_instance_library
(
device_conv2d_bwd_data_instance
${
CONV2D_BWD_DATA_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt
View file @
adfcb029
set
(
DEVICE_CONV2D_FWD_INSTANCES
)
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_CONV2D_FWD_INSTANCES device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_CONV2D_FWD_INSTANCES device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
)
list
(
APPEND DEVICE_CONV2D_FWD_INSTANCES device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"bf16"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_CONV2D_FWD_INSTANCES device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"int8"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_CONV2D_FWD_INSTANCES device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
)
endif
()
list
(
APPEND DEVICE_CONV2D_FWD_INSTANCES device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp
device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
)
add_instance_library
(
device_conv2d_fwd_instance
${
DEVICE_CONV2D_FWD_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/elementwise_normalization/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_elementwise_normalization_instance
device_elementwise_normalization_f16_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt
View file @
adfcb029
set
(
GEMM_INSTANCES
)
if
(
DTYPES MATCHES
"fp64"
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_km_nk_mn_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_km_nk_mn_instance.cpp
)
if
(
DL_KERNELS
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp
)
endif
()
endif
()
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
if
(
DL_KERNELS
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_km_kn_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_km_nk_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dpp_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
)
endif
()
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_add_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_opt_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_add_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_opt_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_add_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_opt_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_add_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_opt_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v2_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_interwave_pipeline_v1_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"int8"
OR NOT DEFINED DTYPES
)
if
(
DL_KERNELS
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_kn_mn_irregular_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_nk_mn_irregular_instance.cpp
)
endif
()
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"bf16"
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_nk_mn_instance.cpp
)
endif
()
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_mk_kn_mn_instance.cpp
device_gemm_xdl_f64_f64_f64_mk_nk_mn_instance.cpp
device_gemm_xdl_f64_f64_f64_km_kn_mn_instance.cpp
device_gemm_xdl_f64_f64_f64_km_nk_mn_instance.cpp
)
add_instance_library
(
device_gemm_instance
${
GEMM_INSTANCES
}
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_mk_kn_mn_instance.cpp
device_gemm_xdl_f32_f32_f32_mk_nk_mn_instance.cpp
device_gemm_xdl_f32_f32_f32_km_kn_mn_instance.cpp
device_gemm_xdl_f32_f32_f32_km_nk_mn_instance.cpp
device_gemm_xdl_c_shuffle_f32_f32_f32_mk_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_f32_f32_f32_mk_nk_mn_instance.cpp
device_gemm_xdl_c_shuffle_f32_f32_f32_km_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_f32_f32_f32_km_nk_mn_instance.cpp
device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp
device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp
device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp
device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp
device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp
device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
device_gemm_dl_f16_f16_f16_km_kn_mn_instance.cpp
device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instance.cpp
device_gemm_dl_f16_f16_f16_km_nk_mn_instance.cpp
device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instance.cpp
device_gemm_dpp_f16_f16_f16_km_kn_mn_instance.cpp
device_gemm_dpp_f16_f16_f16_km_nk_mn_instance.cpp
device_gemm_dpp_f16_f16_f16_mk_kn_mn_instance.cpp
device_gemm_dpp_f16_f16_f16_mk_nk_mn_instance.cpp
device_gemm_dpp_f16_f16_f16_km_kn_mn_irregular_instance.cpp
device_gemm_dpp_f16_f16_f16_km_nk_mn_irregular_instance.cpp
device_gemm_dpp_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
device_gemm_dpp_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
device_gemm_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_add_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_opt_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_interwave_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_interwave_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_add_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_opt_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_interwave_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_interwave_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_add_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_opt_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_interwave_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_interwave_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_add_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_opt_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_interwave_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v1_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v2_instance.cpp
device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_interwave_pipeline_v1_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp
device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instance.cpp
device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp
device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instance.cpp
device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp
device_gemm_dl_i8_i8_i8_km_kn_mn_irregular_instance.cpp
device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp
device_gemm_dl_i8_i8_i8_km_nk_mn_irregular_instance.cpp
device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_i8_i8_i8_mk_nk_mn_instance.cpp
device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp
device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_kn_mn_instance.cpp
device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_nk_mn_instance.cpp
)
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
set
(
ENABLE_PIPELINE_V2_OPT OFF
)
add_instance_library
(
device_gemm_instance
${
GEMM_INSTANCES
}
)
set
(
ENABLE_PIPELINE_V2_OPT OFF
)
if
(
ENABLE_PIPELINE_V2_OPT
)
if
(
ENABLE_PIPELINE_V2_OPT
)
set
(
MAX_ILP_OPTS
-mllvm
-amdgpu-enable-max-ilp-scheduling-strategy
...
...
@@ -137,5 +123,5 @@ if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
set_source_files_properties
(
device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
COMPILE_OPTIONS
"
${
MAX_ILP_OPTS
}
"
COMPILE_DEFINITIONS
"
${
WAVES_PER_EU_DEFS
}
;
${
IGLP_OPT_DEFS
}
"
)
endif
(
ENABLE_PIPELINE_V2_OPT
)
endif
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
endif
(
ENABLE_PIPELINE_V2_OPT
)
library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_gemm_add_add_fastgelu_instance
device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instance.cpp
device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instance.cpp
device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp
device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/gemm_add_fastgelu/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_gemm_add_fastgelu_instance
device_gemm_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp
device_gemm_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp
device_gemm_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instance.cpp
device_gemm_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/gemm_add_relu_add_layernorm/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_gemm_add_relu_add_layernorm_instance
device_gemm_add_relu_add_xdl_c_shuffle_layernorm_f16_km_kn_mn_mn_mn_instance.cpp
device_gemm_add_relu_add_xdl_c_shuffle_layernorm_f16_km_nk_mn_mn_mn_instance.cpp
device_gemm_add_relu_add_xdl_c_shuffle_layernorm_f16_mk_kn_mn_mn_mn_instance.cpp
device_gemm_add_relu_add_xdl_c_shuffle_layernorm_f16_mk_nk_mn_mn_mn_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/gemm_bilinear/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_gemm_bilinear_instance
device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp
device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp
...
...
@@ -9,4 +8,3 @@ add_instance_library(device_gemm_bilinear_instance
device_gemm_bilinear_wmma_c_shuffle_i8_i8_i8_i8_mk_kn_mn_mn_instance.cpp
device_gemm_bilinear_wmma_c_shuffle_i8_i8_i8_i8_mk_nk_mn_mn_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/gemm_fastgelu/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_gemm_fastgelu_instance
device_gemm_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
device_gemm_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
device_gemm_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
device_gemm_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/gemm_multiply_add/CMakeLists.txt
View file @
adfcb029
set
(
GEMM_MULTIPLY_ADD_INSTANCES
)
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_MULTIPLY_ADD_INSTANCES device_gemm_multiply_add_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp
)
list
(
APPEND GEMM_MULTIPLY_ADD_INSTANCES device_gemm_multiply_add_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp
)
endif
()
if
((
DTYPES MATCHES
"fp16"
AND DTYPES MATCHES
"fp8"
)
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_MULTIPLY_ADD_INSTANCES device_gemm_multiply_add_xdl_c_shuffle_f16_f8_f32_f32_f16_mk_kn_mn_mn_mn_instance.cpp
)
list
(
APPEND GEMM_MULTIPLY_ADD_INSTANCES device_gemm_multiply_add_xdl_c_shuffle_f16_f8_f32_f32_f16_mk_nk_mn_mn_mn_instance.cpp
)
endif
()
list
(
APPEND GEMM_MULTIPLY_ADD_INSTANCES device_gemm_multiply_add_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp
device_gemm_multiply_add_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp
device_gemm_multiply_add_xdl_c_shuffle_f16_f8_f32_f32_f16_mk_kn_mn_mn_mn_instance.cpp
device_gemm_multiply_add_xdl_c_shuffle_f16_f8_f32_f32_f16_mk_nk_mn_mn_mn_instance.cpp
)
add_instance_library
(
device_gemm_multiply_add_instance
${
GEMM_MULTIPLY_ADD_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/gemm_splitk/CMakeLists.txt
View file @
adfcb029
set
(
GEMM_SPLITK_INSTANCES
)
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f32_f32_f32_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f32_f32_f32_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f32_f32_f32_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f32_f32_f32_km_nk_mn_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instance.cpp
)
endif
()
if
((
DTYPES MATCHES
"fp16"
AND DTYPES MATCHES
"fp8"
)
OR NOT DEFINED DTYPES
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f8_f16_f16_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f8_f16_f16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f8_f16_f16_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f8_f16_f16_km_nk_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f8_f16_mk_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f8_f16_mk_nk_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f8_f16_km_kn_mn_instance.cpp
)
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f16_f8_f16_km_nk_mn_instance.cpp
)
endif
()
list
(
APPEND GEMM_SPLITK_INSTANCES device_gemm_xdl_splitk_f32_f32_f32_mk_kn_mn_instance.cpp
device_gemm_xdl_splitk_f32_f32_f32_mk_nk_mn_instance.cpp
device_gemm_xdl_splitk_f32_f32_f32_km_kn_mn_instance.cpp
device_gemm_xdl_splitk_f32_f32_f32_km_nk_mn_instance.cpp
device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instance.cpp
device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instance.cpp
device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instance.cpp
device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instance.cpp
device_gemm_xdl_splitk_f8_f16_f16_mk_kn_mn_instance.cpp
device_gemm_xdl_splitk_f8_f16_f16_mk_nk_mn_instance.cpp
device_gemm_xdl_splitk_f8_f16_f16_km_kn_mn_instance.cpp
device_gemm_xdl_splitk_f8_f16_f16_km_nk_mn_instance.cpp
device_gemm_xdl_splitk_f16_f8_f16_mk_kn_mn_instance.cpp
device_gemm_xdl_splitk_f16_f8_f16_mk_nk_mn_instance.cpp
device_gemm_xdl_splitk_f16_f8_f16_km_kn_mn_instance.cpp
device_gemm_xdl_splitk_f16_f8_f16_km_nk_mn_instance.cpp
)
add_instance_library
(
device_gemm_splitk_instance
${
GEMM_SPLITK_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/gemm_streamk/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_gemm_streamk_instance
# device_gemm_xdl_streamk_f32_f32_f32_mk_kn_mn_instance.cpp
# device_gemm_xdl_streamk_f32_f32_f32_mk_nk_mn_instance.cpp
...
...
@@ -9,4 +8,3 @@ add_instance_library(device_gemm_streamk_instance
# device_gemm_xdl_streamk_f16_f16_f16_km_kn_mn_instance.cpp
# device_gemm_xdl_streamk_f16_f16_f16_km_nk_mn_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
View file @
adfcb029
add_instance_library
(
device_grouped_conv3d_fwd_instance
if
((
DTYPES MATCHES
"fp16"
)
OR NOT DEFINED DTYPES
)
device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instance.cpp
device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f16_instance.cpp
device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f16_comp_fp8_instance.cpp
endif
()
if
((
DTYPES MATCHES
"fp32"
)
OR NOT DEFINED DTYPES
)
device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instance.cpp
device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_instance.cpp
endif
()
if
((
DTYPES MATCHES
"bf16"
)
OR NOT DEFINED DTYPES
)
device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instance.cpp
device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_instance.cpp
endif
()
if
((
DTYPES MATCHES
"int8"
)
OR NOT DEFINED DTYPES
)
device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instance.cpp
device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_int8_instance.cpp
endif
()
)
library/src/tensor_operation_instance/gpu/grouped_gemm/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_grouped_gemm_instance
device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp
device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp
...
...
@@ -9,4 +8,3 @@ add_instance_library(device_grouped_gemm_instance
device_grouped_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
device_grouped_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/grouped_gemm_fastgelu/CMakeLists.txt
View file @
adfcb029
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
add_instance_library
(
device_grouped_gemm_fastgelu_instance
device_grouped_gemm_fastgelu_xdl_f16_f16_f16_mk_kn_mn_instance.cpp
device_grouped_gemm_fastgelu_xdl_f16_f16_f16_mk_nk_mn_instance.cpp
device_grouped_gemm_fastgelu_xdl_f16_f16_f16_km_kn_mn_instance.cpp
device_grouped_gemm_fastgelu_xdl_f16_f16_f16_km_nk_mn_instance.cpp
)
endif
()
library/src/tensor_operation_instance/gpu/max_pool_bwd/CMakeLists.txt
View file @
adfcb029
set
(
DEVICE_MAXPOOL_BWD_INSTANCES
)
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_MAXPOOL_BWD_INSTANCES device_max_pool_bwd_f16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"bf16"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_MAXPOOL_BWD_INSTANCES device_max_pool_bwd_bf16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_MAXPOOL_BWD_INSTANCES device_max_pool_bwd_f32_instance.cpp
)
endif
()
list
(
APPEND DEVICE_MAXPOOL_BWD_INSTANCES device_max_pool_bwd_f16_instance.cpp
device_max_pool_bwd_bf16_instance.cpp
device_max_pool_bwd_f32_instance.cpp
)
add_instance_library
(
device_max_pool_bwd_instance
${
DEVICE_MAXPOOL_BWD_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/normalization/CMakeLists.txt
View file @
adfcb029
set
(
DEVICE_NORMALIZATION_INSTANCES
)
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_NORMALIZATION_INSTANCES device_layernorm2d_f16_instance.cpp
list
(
APPEND DEVICE_NORMALIZATION_INSTANCES
device_layernorm2d_f16_instance.cpp
device_layernorm4d_f16_instance.cpp
device_groupnorm_f16_instance.cpp
device_groupnorm_swish_f16_instance.cpp
device_groupnorm_swish_f16_f32_f32_f16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_NORMALIZATION_INSTANCES device_layernorm2d_f32_instance.cpp
device_groupnorm_swish_f16_f32_f32_f16_instance.cpp
device_layernorm2d_f32_instance.cpp
device_layernorm4d_f32_instance.cpp
device_groupnorm_f32_instance.cpp
device_groupnorm_swish_f32_instance.cpp
)
endif
()
add_instance_library
(
device_normalization_instance
${
DEVICE_NORMALIZATION_INSTANCES
}
)
library/src/tensor_operation_instance/gpu/pool3d_fwd/CMakeLists.txt
View file @
adfcb029
set
(
DEVICE_POOL3D_FWD_INSTANCES
)
if
(
DTYPES MATCHES
"fp16"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_POOL3D_FWD_INSTANCES device_avg_pool3d_fwd_ndhwc_f16_instance.cpp
device_max_pool3d_fwd_ndhwc_f16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"bf16"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_POOL3D_FWD_INSTANCES device_avg_pool3d_fwd_ndhwc_bf16_instance.cpp
device_max_pool3d_fwd_ndhwc_bf16_instance.cpp
)
endif
()
if
(
DTYPES MATCHES
"fp32"
OR NOT DEFINED DTYPES
)
list
(
APPEND DEVICE_POOL3D_FWD_INSTANCES device_avg_pool3d_fwd_ndhwc_f32_instance.cpp
device_max_pool3d_fwd_ndhwc_f32_instance.cpp
)
endif
()
list
(
APPEND DEVICE_POOL3D_FWD_INSTANCES device_avg_pool3d_fwd_ndhwc_f16_instance.cpp
device_max_pool3d_fwd_ndhwc_f16_instance.cpp
device_avg_pool3d_fwd_ndhwc_f32_instance.cpp
device_max_pool3d_fwd_ndhwc_f32_instance.cpp
device_avg_pool3d_fwd_ndhwc_bf16_instance.cpp
device_max_pool3d_fwd_ndhwc_bf16_instance.cpp
)
add_instance_library
(
device_pool3d_fwd_instance
${
DEVICE_POOL3D_FWD_INSTANCES
}
)
Prev
1
2
3
4
5
Next
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment
