Integrate universal gemm with conv forward (#1320)

* Integrate universal gemm with conv fwd

* Fix conv fwd wmma test

* Fix instances

* Remove direct load check

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_two_stage_xdl_cshuffle.hpp

@@ -674,7 +674,7 @@ struct DeviceGroupedConvBwdWeightTwoStage_Xdl_CShuffle
                         clear_workspace();
                     };
 
-                    ave_time = ck::utility::launch_and_time_kernel_with_preprocess<false>(
+                    ave_time += ck::utility::launch_and_time_kernel_with_preprocess<false>(
                         stream_config,
                         run_flush_cache,
                         kernel,
@@ -690,7 +690,7 @@ struct DeviceGroupedConvBwdWeightTwoStage_Xdl_CShuffle
                 }
                 else
                 {
-                    ave_time = launch_and_time_kernel_with_preprocess(
+                    ave_time += launch_and_time_kernel_with_preprocess(
                         stream_config,
                         clear_workspace,
                         kernel,

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp

@@ -820,15 +820,7 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle
                 return false;
             }
         }
-        else if(ck::is_lds_direct_load_supported())
-        {
-            if constexpr(!(is_same_v<AccDataType, float> || is_same_v<AccDataType, float> ||
-                           is_same_v<AccDataType, int32_t> || is_same_v<AccDataType, double>))
-            {
-                return false;
-            }
-        }
-        else
+        if(!ck::is_xdl_supported())
         {
             return false;
         }