Merge branch 'd2s-bug' into 'master'

Fix depth_to_space opencl global work size bug.

See merge request !386

mace/kernels/depth_to_space.h

@@ -108,11 +108,11 @@ struct DepthToSpaceOpFunctor<DeviceType::OPENCL, T> {
       : block_size_(block_size), d2s_(d2s) {}
   void operator()(const Tensor *input, Tensor *output, StatsFuture *future);
 
+  const int block_size_;
+  bool d2s_;
   cl::Kernel kernel_;
   uint32_t kwg_size_;
   std::unique_ptr<BufferBase> kernel_error_;
-  const int block_size_;
-  bool d2s_;
   std::vector<index_t> input_shape_;
 };
 

mace/kernels/opencl/depth_to_space_opencl.cc

@@ -22,16 +22,31 @@ void DepthToSpaceOpFunctor<DeviceType::OPENCL, T>::operator()(
 
   const char *kernel_name = nullptr;
 
+  uint32_t gws[3];
+  std::stringstream ss;
   index_t output_height, output_width, output_depth;
-  if (d2s_) { output_height = input_height * block_size_;
+  if (d2s_) {
+    output_height = input_height * block_size_;
     output_width = input_width * block_size_;
     output_depth = input_depth / (block_size_ * block_size_);
     kernel_name = "depth_to_space";
+
+    gws[0] = static_cast<uint32_t>(RoundUpDiv4(output_depth));
+    gws[1] = static_cast<uint32_t>(output_width);
+    gws[2] = static_cast<uint32_t>(output_height * batch);
+    ss << "depth_to_space_opencl_kernel_" << batch << "_"
+       << output_height << "_" << output_width << "_" << output_depth;
   } else {
     output_height = input_height / block_size_;
     output_width = input_width / block_size_;
     output_depth = input_depth * block_size_ * block_size_;
     kernel_name = "space_to_depth";
+
+    gws[0] = static_cast<uint32_t>(RoundUpDiv4(input_depth));
+    gws[1] = static_cast<uint32_t>(input_width);
+    gws[2] = static_cast<uint32_t>(input_height * batch);
+    ss << "space_to_depth_opencl_kernel_" << input->dim(0) << "_"
+       << input->dim(1) << "_" << input->dim(2) << "_" << input->dim(3);
   }
   const index_t input_depth_blocks = RoundUpDiv4(input_depth);
   const index_t output_depth_blocks = RoundUpDiv4(output_depth);
@@ -73,23 +88,7 @@ void DepthToSpaceOpFunctor<DeviceType::OPENCL, T>::operator()(
         static_cast<uint32_t>(runtime->GetKernelMaxWorkGroupSize(kernel_));
   }
 
-  uint32_t gws[3];
-  std::stringstream ss;
   if (!IsVecEqual(input_shape_, input->shape())) {
-    if (d2s_) {
-      gws[0] = static_cast<uint32_t>(output_depth_blocks);
-      gws[1] = static_cast<uint32_t>(output_width);
-      gws[2] = static_cast<uint32_t>(output_height * batch);
-      ss << "depth_to_space_opencl_kernel_" << output->dim(0) << "_"
-         << output->dim(1) << "_" << output->dim(2) << "_" << output->dim(3);
-    } else {
-      gws[0] = static_cast<uint32_t>(input_depth_blocks);
-      gws[1] = static_cast<uint32_t>(input_width);
-      gws[2] = static_cast<uint32_t>(input_height * batch);
-      ss << "space_to_depth_opencl_kernel_" << input->dim(0) << "_"
-         << input->dim(1) << "_" << input->dim(2) << "_" << input->dim(3);
-    }
-
     uint32_t idx = 0;
     if (runtime->IsOutOfRangeCheckEnabled()) {
       kernel_.setArg(idx++,