fix bug of KernelDepthwiseConvNHWC

Author: OuyangChao

paddle/fluid/operators/math/depthwise_conv.cu

@@ -142,7 +142,7 @@ __device__ __inline__ void KernelDepthwiseConvNHWC(
     for (int w_in = w_in_start; w_in < w_in_end; w_in += dilate_width) {
       if (h_in >= h_start && h_in < h_end && w_in >= w_start && w_in < w_end) {
         int offset = ((batch * input_height + h_in) * input_width + w_in) *
-                         output_channels +
+                         input_channels +
                      c_in;
         T in_data = input_data[offset];
         if (fuse_relu_before_conv) {

python/paddle/fluid/tests/unittests/test_conv2d_op.py

@@ -1235,6 +1235,7 @@ def init_paddings(self):
 create_test_channel_last_class(TestWithPad_AsyPadding)
 create_test_channel_last_class(TestWithGroup_AsyPadding)
 create_test_channel_last_class(TestWith1x1_AsyPadding)
+create_test_channel_last_class(TestWithDepthWise7x7_AsyPadding)
 create_test_channel_last_class(TestWithInput1x1Filter1x1_AsyPadding)
 
 create_test_channel_last_class(TestDepthwiseConv_AsyPadding)