merge haocong branch

library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp

@@ -22,7 +22,6 @@ using I8   = int8_t;
 using I32  = int32_t;
 using F8   = ck::f8_t;
 using BF8  = ck::bf8_t;
-using I4   = ck::pk_i4_t;
 
 using Empty_Tuple = ck::Tuple<>;
 

library/include/ck/library/tensor_operation_instance/gpu/gemm_dl.inc

@@ -28,6 +28,16 @@ void add_device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instances(
         DeviceGemm<Col, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
         instances);
 
+void add_device_gemm_dpp_f16_f16_f16_km_kn_mn_instances(
+    std::vector<std::unique_ptr<
+        DeviceGemm<Col, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_gemm_dpp_f16_f16_f16_km_kn_mn_irregular_instances(
+    std::vector<std::unique_ptr<
+        DeviceGemm<Col, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
 void add_device_gemm_dl_f16_f16_f16_km_nk_mn_instances(
     std::vector<std::unique_ptr<
         DeviceGemm<Col, Col, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
@@ -38,6 +48,16 @@ void add_device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instances(
         DeviceGemm<Col, Col, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
         instances);
 
+void add_device_gemm_dpp_f16_f16_f16_km_nk_mn_instances(
+    std::vector<std::unique_ptr<
+        DeviceGemm<Col, Col, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_gemm_dpp_f16_f16_f16_km_nk_mn_irregular_instances(
+    std::vector<std::unique_ptr<
+        DeviceGemm<Col, Col, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
 void add_device_gemm_dl_f16_f16_f16_mk_kn_mn_instances(
     std::vector<std::unique_ptr<
         DeviceGemm<Row, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
@@ -48,6 +68,16 @@ void add_device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instances(
         DeviceGemm<Row, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
         instances);
 
+void add_device_gemm_dpp_f16_f16_f16_mk_kn_mn_instances(
+    std::vector<std::unique_ptr<
+        DeviceGemm<Row, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_gemm_dpp_f16_f16_f16_mk_kn_mn_irregular_instances(
+    std::vector<std::unique_ptr<
+        DeviceGemm<Row, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
 void add_device_gemm_dl_f16_f16_f16_mk_nk_mn_instances(
     std::vector<std::unique_ptr<
         DeviceGemm<Row, Col, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&

library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp

@@ -304,23 +304,7 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGroupe
                     op_ptrs);
             }
 #endif
-#ifdef CK_ENABLE_BF16
-            if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
-                         is_same_v<WeiDataType, ck::bhalf_t> &&
-                         is_same_v<OutDataType, ck::bhalf_t> &&
-                         is_same_v<AComputeType, ck::bhalf_t> &&
-                         is_same_v<BComputeType, ck::bhalf_t>)
-            {
-                add_device_grouped_conv2d_fwd_xdl_merged_groups_ngchw_gkyxc_ngkhw_bf16_instances(
-                    op_ptrs);
-                add_device_grouped_conv2d_fwd_xdl_ngchw_gkyxc_ngkhw_bf16_instances(op_ptrs);
-                add_device_grouped_conv2d_fwd_xdl_ngchw_gkyxc_ngkhw_bf16_comp_instances(op_ptrs);
-                add_device_grouped_conv2d_fwd_xdl_ngchw_gkyxc_ngkhw_bf16_mem_intra_instances(
-                    op_ptrs);
-                add_device_grouped_conv2d_fwd_xdl_ngchw_gkyxc_ngkhw_bf16_mem_inter_instances(
-                    op_ptrs);
-            }
-#endif
+
 #ifdef CK_ENABLE_INT8
             if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
                          is_same_v<OutDataType, int8_t> && is_same_v<AComputeType, int8_t> &&