Skip to content

GitLab

Commit fcb9e1b4 authored by M.Emin Ozturk's avatar M.Emin Ozturk
Browse files

clang all

parent bd0a6f73
Hide whitespace changes
Inline Side-by-side
Showing with 775 additions and 766 deletions
library/include/ck/library/tensor_operation_instance/gpu/gemm_universal_streamk.hpp 100755 → 100644
View file @ fcb9e1b4
...@@ -17,216 +17,216 @@ namespace tensor_operation { ...@@ -17,216 +17,216 @@ namespace tensor_operation {
namespace device { namespace device {
namespace instance { namespace instance {
#ifdef CK_ENABLE_FP16 #ifdef CK_ENABLE_FP16
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v1_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v2_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Row, Row,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v1_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v2_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
F16, F16,
F16, F16,
F16, F16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>& instances); PassThrough>>>& instances);
void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
...@@ -240,382 +240,381 @@ namespace instance { ...@@ -240,382 +240,381 @@ namespace instance {
// Emin @Added // Emin @Added
#ifdef CK_ENABLE_BF16 #ifdef CK_ENABLE_BF16
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_kn_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Row, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_mkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_mkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_mkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_mkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& DeviceGemmV2<Col, Col, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
#endif #endif
// Emin @Added // Emin @Added
#if(defined(CK_ENABLE_FP8)) #if(defined(CK_ENABLE_FP8))
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_kn_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f16_f8_f16_mk_nk_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F16, F8, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_kn_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Row, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v1_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v2_default_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_f8_f16_f16_mk_nk_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr< std::vector<std::unique_ptr<
DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>& DeviceGemm_Streamk_V2<Row, Col, Row, F8, F16, F16, PassThrough, PassThrough, PassThrough>>>&
instances); instances);
...@@ -711,7 +710,7 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemm_S ...@@ -711,7 +710,7 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemm_S
} }
#endif #endif
//Emin @Added // Emin @Added
#ifdef CK_ENABLE_BF16 #ifdef CK_ENABLE_BF16
if constexpr(is_same_v<ADataType, bhalf_t> && is_same_v<BDataType, bhalf_t> && if constexpr(is_same_v<ADataType, bhalf_t> && is_same_v<BDataType, bhalf_t> &&
is_same_v<CDataType, bhalf_t>) is_same_v<CDataType, bhalf_t>)
...@@ -814,7 +813,7 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemm_S ...@@ -814,7 +813,7 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemm_S
} }
} }
#endif #endif
//EMin @Added // EMin @Added
#if(defined(CK_ENABLE_FP8)) #if(defined(CK_ENABLE_FP8))
if constexpr(is_same_v<ADataType, half_t> && is_same_v<BDataType, f8_t> && if constexpr(is_same_v<ADataType, half_t> && is_same_v<BDataType, f8_t> &&
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_bf16_bf16_bf16_mk_nk_mn_comp_default_instance.cpp
View file @ fcb9e1b4
...@@ -9,18 +9,19 @@ namespace device { ...@@ -9,18 +9,19 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Row,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_instances<GemmDefault>{}); instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_comp_instances<GemmDefault>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_default_instance.cpp
View file @ fcb9e1b4
...@@ -9,18 +9,18 @@ namespace device { ...@@ -9,18 +9,18 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_instances<GemmDefault>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_instances<GemmDefault>{});
} }
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_kpadding_instance.cpp
View file @ fcb9e1b4
...@@ -10,14 +10,14 @@ namespace instance { ...@@ -10,14 +10,14 @@ namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnkpadding_instance.cpp
View file @ fcb9e1b4
...@@ -10,14 +10,14 @@ namespace instance { ...@@ -10,14 +10,14 @@ namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,15 +9,15 @@ namespace device { ...@@ -9,15 +9,15 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_default_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Intrawave, GemmDefault>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Intrawave,
GemmDefault>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_kpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Intrawave, GemmKPadding>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Intrawave,
GemmKPadding>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_mnkpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v1_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Intrawave,GemmMNKPadding>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Intrawave,
GemmMNKPadding>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_default_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Interwave, GemmDefault>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Interwave,
GemmDefault>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_kpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Interwave, GemmKPadding>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Interwave,
GemmKPadding>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_mnkpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,20 +9,20 @@ namespace device { ...@@ -9,20 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_mnkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_v2_mnkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Row, Row,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Interwave, device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_mem_instances<Interwave,
GemmMNKPadding>{}); GemmMNKPadding>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_default_instance.cpp
View file @ fcb9e1b4
...@@ -9,18 +9,19 @@ namespace device { ...@@ -9,18 +9,19 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_instances<GemmDefault>{}); instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_instances<GemmDefault>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_kpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,15 +9,15 @@ namespace device { ...@@ -9,15 +9,15 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mkpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,15 +9,15 @@ namespace device { ...@@ -9,15 +9,15 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,15 +9,15 @@ namespace device { ...@@ -9,15 +9,15 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_comp_mpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_default_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Intrawave, GemmDefault>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Intrawave,
GemmDefault>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_kpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_kpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_kpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Intrawave, GemmKPadding>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Intrawave,
GemmKPadding>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_mkpadding_instance.cpp
View file @ fcb9e1b4
...@@ -9,20 +9,20 @@ namespace device { ...@@ -9,20 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_mkpadding_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v1_mkpadding_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Intrawave, device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Intrawave,
GemmMKPadding>{}); GemmMKPadding>{});
} }
} // namespace instance } // namespace instance
......
library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_default_instance.cpp
View file @ fcb9e1b4
...@@ -9,19 +9,20 @@ namespace device { ...@@ -9,19 +9,20 @@ namespace device {
namespace instance { namespace instance {
void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_default_instances( void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_v2_default_instances(
std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
Col, Col,
Row, Row,
BF16, BF16,
BF16, BF16,
BF16, BF16,
PassThrough, PassThrough,
PassThrough, PassThrough,
PassThrough>>>&instances) PassThrough>>>& instances)
{ {
add_device_operation_instances( add_device_operation_instances(
instances, instances,
device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Interwave, GemmDefault>{}); device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_nk_mn_mem_instances<Interwave,
GemmDefault>{});
} }
} // namespace instance } // namespace instance
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment