run simple example

device_operation/include/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp

@@ -465,7 +465,7 @@ struct DeviceConv2dWrWXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_W
     static bool IsSupportedArgument(const Argument& arg)
     {
         // vector load A/B matrix from global memory
-        if(!(ABlockTransferSrcVectorDim == 1 && BBlockTransferSrcVectorDim == 2 &&
+        if(!(ABlockTransferSrcVectorDim == 1 && BBlockTransferSrcVectorDim == 1 &&
              arg.Conv_K_ % ABlockTransferSrcScalarPerVector == 0 &&
              arg.Conv_C_ % BBlockTransferSrcScalarPerVector == 0))
         {

example/14_conv2d_backward_weight_xdl/main.cpp

@@ -51,18 +51,18 @@ using DeviceConvWrWInstance = ck::tensor_operation::device::
         2,                                // MXdlPerWave
         2,                                // NXdlPerWave
         S<4, 16, 4>,                      // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<0, 2, 1>,                       // ABlockTransferThreadClusterArrangeOrder
+        S<2, 0, 1>,                       // ABlockTransferThreadClusterArrangeOrder
         S<1, 0, 2>,                       // ABlockTransferSrcAccessOrder
         1,                                // ABlockTransferSrcVectorDim
         8,                                // ABlockTransferSrcScalarPerVector
-        8,                                // ABlockTransferDstScalarPerVector_K1
+        2,                                // ABlockTransferDstScalarPerVector_K1
         true,                             // ABlockLdsAddExtraM
         S<4, 16, 4>,                      // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,                       // BBlockTransferThreadClusterArrangeOrder
+        S<2, 0, 1>,                       // BBlockTransferThreadClusterArrangeOrder
         S<1, 0, 2>,                       // BBlockTransferSrcAccessOrder
-        2,                                // BBlockTransferSrcVectorDim
+        1,                                // BBlockTransferSrcVectorDim
         8,                                // BBlockTransferSrcScalarPerVector
-        8,                                // BBlockTransferDstScalarPerVector_K1
+        2,                                // BBlockTransferDstScalarPerVector_K1
         true,                             // BBlockLdsAddExtraN
         1,                                // CShuffleMXdlPerWavePerShuffle
         1,                                // CShuffleNXdlPerWavePerShuffle