diff --git a/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp b/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp
index 9efcc5d8c8..682aba0860 100644
--- a/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp
+++ b/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp
@@ -16,6 +16,7 @@
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
 #include "ck/host_utility/device_prop.hpp"
 #include "ck/host_utility/kernel_launch.hpp"
+#include "ck/host_utility/io.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -796,7 +797,7 @@ struct DeviceGroupedConvBwdDataMultipleD_Xdl_CShuffle_v1
             // check if it's 1x1, stride=1 pad = 0 conv
             for(int i = 0; i < NDimSpatial; i++)
             {
-                if(!(arg.filter_spatial_lengths_[i] == 1 && arg.conv_filter_strides_[i] == 1 &&
+                if(!(arg.b_g_k_c_xs_lengths_[3 + i] == 1 && arg.conv_filter_strides_[i] == 1 &&
                      arg.input_left_pads_[i] == 0 && arg.input_right_pads_[i] == 0))
                 {
                     return false;