COMPMID-1145: (API) Introduce prepare() stage (NEON/CL/GLES)

Change-Id: I5b46764f9c3154ec3e3b9c951cc9e6dfbcb81dfb
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/134255
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Pablo Tello <pablo.tello@arm.com>
Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com>

1 files changed, 19 insertions, 1 deletions

diff --git a/src/runtime/CL/functions/CLRNNLayer.cpp b/src/runtime/CL/functions/CLRNNLayer.cpp
index 4843ba6364..0e1b9d5b58 100644
--- a/src/runtime/CL/functions/CLRNNLayer.cpp
+++ b/src/runtime/CL/functions/CLRNNLayer.cpp
@@ -36,7 +36,8 @@ using namespace arm_compute;
 using namespace arm_compute::misc::shape_calculator;
 
 CLRNNLayer::CLRNNLayer(std::shared_ptr<IMemoryManager> memory_manager)
-    : _memory_group(std::move(memory_manager)), _gemm_state_f(), _add_kernel(), _activation_kernel(), _fully_connected_kernel(), _copy_kernel(), _fully_connected_out(), _gemm_output(), _add_output()
+    : _memory_group(std::move(memory_manager)), _gemm_state_f(), _add_kernel(), _activation_kernel(), _fully_connected_kernel(), _copy_kernel(), _fully_connected_out(), _gemm_output(), _add_output(),
+      _is_prepared(false)
 {
 }
 
@@ -74,6 +75,8 @@ void CLRNNLayer::configure(const ICLTensor *input, const ICLTensor *weights, con
     const int   idx_height = get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::HEIGHT);
     TensorShape shape      = compute_rnn_shape(recurrent_weights->info(), hidden_state->info()->dimension(idx_height));
 
+    _is_prepared = false;
+
     _fully_connected_out.allocator()->init(TensorInfo(shape, 1, input->info()->data_type()));
     _gemm_output.allocator()->init(TensorInfo(shape, 1, input->info()->data_type()));
 
@@ -100,7 +103,10 @@ void CLRNNLayer::configure(const ICLTensor *input, const ICLTensor *weights, con
 
 void CLRNNLayer::run()
 {
+    prepare();
+
     _memory_group.acquire();
+
     _fully_connected_kernel.run();
     _gemm_state_f.run();
     CLScheduler::get().enqueue(_add_kernel);
@@ -108,5 +114,17 @@ void CLRNNLayer::run()
 
     // copy hidden out to output
     CLScheduler::get().enqueue(_copy_kernel);
+
     _memory_group.release();
+}
+
+void CLRNNLayer::prepare()
+{
+    if(!_is_prepared)
+    {
+        _fully_connected_kernel.prepare();
+        _gemm_state_f.prepare();
+
+        _is_prepared = true;
+    }
 }
\ No newline at end of file