1 files changed, 21 insertions, 13 deletions

diff --git a/src/runtime/CL/functions/CLStackLayer.cpp b/src/runtime/CL/functions/CLStackLayer.cpp
index 79c3fe5371..c15496fc31 100644
--- a/src/runtime/CL/functions/CLStackLayer.cpp
+++ b/src/runtime/CL/functions/CLStackLayer.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2020 ARM Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -21,8 +21,6 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#include <complex>
-
 #include "arm_compute/runtime/CL/functions/CLStackLayer.h"
 
 #include "arm_compute/core/CL/ICLTensor.h"
@@ -33,31 +31,41 @@
 #include "arm_compute/core/utils/misc/ShapeCalculator.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
 
+#include "src/common/utils/Log.h"
+#include "src/core/CL/kernels/CLStackLayerKernel.h"
+
+#include <complex>
+
 namespace arm_compute
 {
 CLStackLayer::CLStackLayer() // NOLINT
-    : _input(),
-      _stack_kernels(),
-      _num_inputs(0)
+    : _input(), _stack_kernels(), _num_inputs(0)
 {
 }
 
+CLStackLayer::~CLStackLayer() = default;
+
 void CLStackLayer::configure(const std::vector<ICLTensor *> &input, int axis, ICLTensor *output)
 {
     configure(CLKernelLibrary::get().get_compile_context(), input, axis, output);
 }
 
-void CLStackLayer::configure(const CLCompileContext &compile_context, const std::vector<ICLTensor *> &input, int axis, ICLTensor *output)
+void CLStackLayer::configure(const CLCompileContext         &compile_context,
+                             const std::vector<ICLTensor *> &input,
+                             int                             axis,
+                             ICLTensor                      *output)
 {
+    ARM_COMPUTE_LOG_PARAMS(input, axis, output);
     _num_inputs = input.size();
-    _stack_kernels.resize(_num_inputs);
+    _stack_kernels.reserve(_num_inputs);
 
     // Wrap around negative values
     const unsigned int axis_u = wrap_around(axis, static_cast<int>(input[0]->info()->num_dimensions() + 1));
 
-    for(unsigned int i = 0; i < _num_inputs; i++)
+    for (unsigned int i = 0; i < _num_inputs; i++)
     {
-        _stack_kernels[i].configure(compile_context, input[i], axis_u, i, _num_inputs, output);
+        _stack_kernels.emplace_back(std::make_unique<CLStackLayerKernel>());
+        _stack_kernels.back()->configure(compile_context, input[i], axis_u, i, _num_inputs, output);
     }
 }
 
@@ -72,7 +80,7 @@ Status CLStackLayer::validate(const std::vector<ITensorInfo *> &input, int axis,
 
     const unsigned int num_inputs = input.size();
 
-    for(unsigned int i = 0; i < num_inputs; i++)
+    for (unsigned int i = 0; i < num_inputs; i++)
     {
         // All the tensors must have the same rank
         ARM_COMPUTE_RETURN_ERROR_ON(input[i]->num_dimensions() != rank);
@@ -85,9 +93,9 @@ Status CLStackLayer::validate(const std::vector<ITensorInfo *> &input, int axis,
 
 void CLStackLayer::run()
 {
-    for(unsigned i = 0; i < _num_inputs; i++)
+    for (unsigned i = 0; i < _num_inputs; i++)
     {
-        CLScheduler::get().enqueue(_stack_kernels[i], false);
+        CLScheduler::get().enqueue(*_stack_kernels[i], false);
     }
 }
 } // namespace arm_compute