diff options
Diffstat (limited to 'arm_compute/runtime/NEON/functions/NEConcatenateLayer.h')
| -rw-r--r-- | arm_compute/runtime/NEON/functions/NEConcatenateLayer.h | 22 |
1 files changed, 15 insertions, 7 deletions
diff --git a/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h b/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h index 2cdc720fb6..b70d6ebc7c 100644 --- a/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h +++ b/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018 ARM Limited. + * Copyright (c) 2018-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -26,6 +26,7 @@ #include "arm_compute/runtime/IFunction.h" +#include "arm_compute/core/NEON/kernels/NEHeightConcatenateLayerKernel.h" #include "arm_compute/core/Types.h" #include <memory> @@ -41,6 +42,7 @@ class Status; /** Basic function to execute concatenate tensors along a given axis. This function calls the following kernels: * * -# @ref NEWidthConcatenateLayer (if underlying concatenation axis is 0). + * -# @ref NEHeightConcatenateLayerKernel (if underlying concatenation axis is 1). * -# @ref NEDepthConcatenateLayer (if underlying concatenation axis is 2). */ class NEConcatenateLayer : public IFunction @@ -51,21 +53,21 @@ public: /** Initialise the kernel's inputs vector and output. * * @note Input and output tensor dimensions preconditions defer depending on the concatenation axis. - * @note Preconditions can be found respectively at @ref NEWidthConcatenateLayer and @ref NEDepthConcatenateLayer. + * @note Preconditions can be found respectively at @ref NEWidthConcatenateLayer, @ref NEHeightConcatenateLayerKernel and @ref NEDepthConcatenateLayer. * * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: QASYMM8/F16/F32. * @param[out] output Output tensor. Data types supported: Same as @p input. - * @param[in] axis Concatenation axis. Supported underlying concatenation axis are 0 and 2. + * @param[in] axis Concatenation axis. Supported underlying concatenation axis are 0, 1 and 2. */ void configure(const std::vector<ITensor *> &inputs_vector, ITensor *output, DataLayoutDimension axis); /** Static function to check if given info will lead to a valid configuration of @ref NEConcatenateLayer * * @note Input and output tensor dimensions preconditions defer depending on the concatenation axis. - * @note Preconditions can be found respectively at @ref NEWidthConcatenateLayer and @ref NEDepthConcatenateLayer. + * @note Preconditions can be found respectively at @ref NEWidthConcatenateLayer, @ref NEHeightConcatenateLayerKernel and @ref NEDepthConcatenateLayer. * * @param[in] inputs_vector The vectors containing all the tensors info to concatenate. Data types supported: QASYMM8/F16/F32. * @param[in] output Output tensor info. Data types supported: Same as @p input. - * @param[in] axis Concatenation axis. Supported underlying concatenation axis are 0 and 2. + * @param[in] axis Concatenation axis. Supported underlying concatenation axis are 0, 1 and 2. * * @return a status */ @@ -75,7 +77,13 @@ public: void run() override; private: - std::unique_ptr<IFunction> _concat_function; + void configure_h_concatenate(std::vector<ITensor *> inputs_vector, ITensor *output); + static Status validate_h_concatenate(const std::vector<ITensorInfo *> &inputs_vector, const ITensorInfo *output); + + std::unique_ptr<IFunction> _concat_function; + std::unique_ptr<NEHeightConcatenateLayerKernel[]> _hconcat_kernels; + unsigned int _num_inputs; + unsigned int _axis; }; -} +} // namespace arm_compute #endif /* __ARM_COMPUTE_NECONCATENATELAYER_H__ */ |