dwconv5x5s1 pack4 bf16s neon kernel

src/layer/arm/convolutiondepthwise_arm.cpp

@@ -555,15 +555,15 @@ int ConvolutionDepthWise_arm::forward_bf16s(const Mat& bottom_blob, Mat& top_blo
                     activation->forward_inplace(top_blob, opt);
                 }
             }
-//             else if (kernel_w == 5 && kernel_h == 5 && dilation_w == 1 && dilation_h == 1 && stride_w == 1 && stride_h == 1)
-//             {
-//                 convdw5x5s1_pack4_bf16s_neon(bottom_blob_bordered, top_blob, weight_data_pack4_bf16, bias_data, opt);
-//
-//                 if (activation)
-//                 {
-//                     activation->forward_inplace(top_blob, opt);
-//                 }
-//             }
+            else if (kernel_w == 5 && kernel_h == 5 && dilation_w == 1 && dilation_h == 1 && stride_w == 1 && stride_h == 1)
+            {
+                convdw5x5s1_pack4_bf16s_neon(bottom_blob_bordered, top_blob, weight_data_pack4_bf16, bias_data, opt);
+
+                if (activation)
+                {
+                    activation->forward_inplace(top_blob, opt);
+                }
+            }
             else if (kernel_w == 5 && kernel_h == 5 && dilation_w == 1 && dilation_h == 1 && stride_w == 2 && stride_h == 2)
             {
                 convdw5x5s2_pack4_bf16s_neon(bottom_blob_bordered, top_blob, weight_data_pack4_bf16, bias_data, opt);