COMPMID-3388: Async support to CLReshapeLayerKernel kernels/functions

Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Change-Id: I141a943dfd691069317860e852ecdd0ba7391604
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3501
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

5 files changed, 60 insertions, 12 deletions

diff --git a/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h b/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h
index 997bb79a47..64b7ea6668 100644
--- a/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h
+++ b/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h
@@ -25,9 +25,9 @@
 #define ARM_COMPUTE_CLARGMINMAXLAYER_H
 
 #include "arm_compute/core/CL/kernels/CLArgMinMaxLayerKernel.h"
-#include "arm_compute/core/CL/kernels/CLReshapeLayerKernel.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/functions/CLReshapeLayer.h"
 #include "arm_compute/runtime/IFunction.h"
 #include "arm_compute/runtime/IMemoryManager.h"
 #include "arm_compute/runtime/MemoryGroup.h"
@@ -89,7 +89,7 @@ private:
     std::vector<CLTensor>               _results_vector;
     CLTensor                            _not_reshaped_output;
     std::vector<CLArgMinMaxLayerKernel> _reduction_kernels_vector;
-    CLReshapeLayerKernel                _reshape_kernel;
+    CLReshapeLayer                      _reshape;
     unsigned int                        _num_of_stages;
     unsigned int                        _reduction_axis;
 };
diff --git a/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h b/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h
index 91b30fabcb..490b07bfe4 100644
--- a/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h
+++ b/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h
@@ -29,10 +29,10 @@
 #include "arm_compute/core/CL/kernels/CLPadLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLPermuteKernel.h"
 #include "arm_compute/core/CL/kernels/CLQuantizationLayerKernel.h"
-#include "arm_compute/core/CL/kernels/CLReshapeLayerKernel.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
 #include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/functions/CLReshapeLayer.h"
 #include "arm_compute/runtime/CPP/CPPScheduler.h"
 #include "arm_compute/runtime/CPP/functions/CPPBoxWithNonMaximaSuppressionLimit.h"
 #include "arm_compute/runtime/IFunction.h"
@@ -131,9 +131,9 @@ private:
 
     // OpenCL kernels
     CLPermuteKernel              _permute_deltas_kernel;
-    CLReshapeLayerKernel         _flatten_deltas_kernel;
+    CLReshapeLayer               _flatten_deltas;
     CLPermuteKernel              _permute_scores_kernel;
-    CLReshapeLayerKernel         _flatten_scores_kernel;
+    CLReshapeLayer               _flatten_scores;
     CLComputeAllAnchorsKernel    _compute_anchors_kernel;
     CLBoundingBoxTransformKernel _bounding_box_kernel;
     CLPadLayerKernel             _pad_kernel;
diff --git a/arm_compute/runtime/CL/functions/CLReductionOperation.h b/arm_compute/runtime/CL/functions/CLReductionOperation.h
index 25cf655802..b420bbf631 100644
--- a/arm_compute/runtime/CL/functions/CLReductionOperation.h
+++ b/arm_compute/runtime/CL/functions/CLReductionOperation.h
@@ -26,9 +26,9 @@
 
 #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h"
 #include "arm_compute/core/CL/kernels/CLReductionOperationKernel.h"
-#include "arm_compute/core/CL/kernels/CLReshapeLayerKernel.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/functions/CLReshapeLayer.h"
 #include "arm_compute/runtime/IFunction.h"
 #include "arm_compute/runtime/IMemoryManager.h"
 #include "arm_compute/runtime/MemoryGroup.h"
@@ -94,7 +94,7 @@ private:
     std::vector<CLTensor>                   _results_vector;
     std::vector<CLReductionOperationKernel> _reduction_kernels_vector;
     std::vector<CLFillBorderKernel>         _border_handlers_vector;
-    CLReshapeLayerKernel                    _reshape_kernel;
+    CLReshapeLayer                          _reshape;
     ReductionOperation                      _op;
     unsigned int                            _num_of_stages;
     unsigned int                            _reduction_axis;
diff --git a/arm_compute/runtime/CL/functions/CLReshapeLayer.h b/arm_compute/runtime/CL/functions/CLReshapeLayer.h
index e91c2c739b..9017384545 100644
--- a/arm_compute/runtime/CL/functions/CLReshapeLayer.h
+++ b/arm_compute/runtime/CL/functions/CLReshapeLayer.h
@@ -24,6 +24,7 @@
 #ifndef ARM_COMPUTE_CLRESHAPELAYER_H
 #define ARM_COMPUTE_CLRESHAPELAYER_H
 
+#include "arm_compute/runtime/CL/ICLOperator.h"
 #include "arm_compute/runtime/CL/ICLSimpleFunction.h"
 
 namespace arm_compute
@@ -31,9 +32,21 @@ namespace arm_compute
 class ICLTensor;
 
 /** Basic function to run @ref CLReshapeLayerKernel */
-class CLReshapeLayer : public ICLSimpleFunction
+class CLReshapeLayer : public IFunction
 {
 public:
+    /** Default Constructor */
+    CLReshapeLayer();
+    /** Default Destructor */
+    ~CLReshapeLayer();
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    CLReshapeLayer(const CLReshapeLayer &) = delete;
+    /** Default move constructor */
+    CLReshapeLayer(CLReshapeLayer &&);
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    CLReshapeLayer &operator=(const CLReshapeLayer &) = delete;
+    /** Default move assignment operator */
+    CLReshapeLayer &operator=(CLReshapeLayer &&);
     /** Initialise the kernel's inputs and outputs
      *
      * @param[in]  input  First tensor input. Data type supported: All
@@ -56,6 +69,41 @@ public:
      * @return a status
      */
     static Status validate(const ITensorInfo *input, const ITensorInfo *output);
+
+    // Inherited methods overridden:
+    void run() override;
+
+private:
+    struct Impl;
+    std::unique_ptr<Impl> _impl;
+};
+
+namespace experimental
+{
+/** Basic function to run @ref CLReshapeLayerKernel */
+class CLReshapeLayer : public ICLOperator
+{
+public:
+    /** Initialise the kernel's inputs and outputs
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  input           Input tensor info. Data type supported: All
+     * @param[out] output          Output info. Data type supported: Same as @p input
+     */
+    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
+
+    /** Static function to check if given info will lead to a valid configuration of @ref CLReshapeLayer
+     *
+     * @param[in] input  Input tensor info. Data type supported: All
+     * @param[in] output Output tensor info. Data type supported: Same as @p input
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
+
+    // Inherited methods overridden:
+    MemoryRequirements workspace() const override;
 };
+} // namespace experimental
 }
 #endif /*ARM_COMPUTE_CLRESHAPELAYER_H */
diff --git a/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h b/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h
index 231a56f712..f0ef15acd7 100644
--- a/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h
+++ b/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h
@@ -24,10 +24,10 @@
 #ifndef ARM_COMPUTE_CLSOFTMAXLAYER_H
 #define ARM_COMPUTE_CLSOFTMAXLAYER_H
 
-#include "arm_compute/core/CL/kernels/CLFlattenLayerKernel.h"
-#include "arm_compute/core/CL/kernels/CLReshapeLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLSoftmaxLayerKernel.h"
 #include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/functions/CLFlattenLayer.h"
+#include "arm_compute/runtime/CL/functions/CLReshapeLayer.h"
 #include "arm_compute/runtime/IFunction.h"
 #include "arm_compute/runtime/IMemoryManager.h"
 #include "arm_compute/runtime/MemoryGroup.h"
@@ -136,8 +136,8 @@ private:
     MemoryGroup                    _memory_group;
     CLLogits1DMaxShiftExpSumKernel _max_shift_exp_sum_kernel;
     CLLogits1DNormKernel           _norm_kernel;
-    std::unique_ptr<ICLKernel>     _flatten_kernel_ptr;
-    CLReshapeLayerKernel           _reshape_kernel;
+    std::unique_ptr<IFunction>     _flatten_ptr;
+    CLReshapeLayer                 _reshape;
     CLTensor                       _max;
     CLTensor                       _sum;
     CLTensor                       _tmp;