diff options
Diffstat (limited to 'compute/ARMComputeEx/arm_compute/runtime')
8 files changed, 146 insertions, 133 deletions
diff --git a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLDirectTransposeConvLayer.h b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLDirectTransposeConvLayer.h index 409eaf593..026209f69 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLDirectTransposeConvLayer.h +++ b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLDirectTransposeConvLayer.h @@ -106,22 +106,24 @@ public: CLDirectTransposeConvLayer &operator=(CLDirectTransposeConvLayer &&) = default; /** Set the input, weights, biases and output tensors. * - * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and an - * optional 4th dimension for batch of inputs. - * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. - * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data type - * supported: Same as @p input. - * @param[in] bias (Optional) The biases have one dimension. - * Data type supported: Should match @p input data type, except for - * input of QASYMM8 and QASYMM8_SIGNED type where biases should be of S32 type - * @param[out] output Output tensor. The output has the same number of dimensions as the - * @p input. - * @param[in] info Contains padding and policies to be used in the deconvolution, this - * is decribed in @ref PadStrideInfo. - * @param[in] invalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. - * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, - * specifies if the weights tensor has been reshaped with @ref CLWeightsReshapeKernel. + * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. + * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. + * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. + * @param[in] bias (Optional) The biases have one dimension. + * Data type supported: Should match @p input data type, + * except for input of QASYMM8 and QASYMM8_SIGNED type + * where biases should be of S32 type + * @param[out] output Output tensor. + * The output has the same number of dimensions as the @p input. + * @param[in] info Contains padding and policies to be used in the deconvolution, + * this is decribed in @ref PadStrideInfo. + * @param[in] invalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in] weights_info (Optional) Weights information needed for + * @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with + * @ref CLWeightsReshapeKernel. * */ void configure(ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, @@ -130,23 +132,24 @@ public: /** Set the input, weights, biases and output tensors. * * @param[in] compile_context The compile context to be used. - * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and - * an optional 4th dimension for batch of inputs. + * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. - * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data - * type supported: Same as @p input. + * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. * @param[in] bias (Optional) The biases have one dimension. * Data type supported: Should match @p input data type, except for - * input of QASYMM8 and QASYMM8_SIGNED type where biases should be of S32 type + * input of QASYMM8 and QASYMM8_SIGNED type + * where biases should be of S32 type * @param[out] output Output tensor. The output has the same number of dimensions as - * the @p input. + * the @p input. * @param[in] info Contains padding and policies to be used in the deconvolution, - * this is decribed in @ref PadStrideInfo. - * @param[in] invalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. - * @param[in] weights_info (Optional) Weights information needed for @ref - * CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref - * CLWeightsReshapeKernel. + * this is decribed in @ref PadStrideInfo. + * @param[in] invalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in] weights_info (Optional) Weights information needed for + * @ref CLConvolutionLayer, specifies if the weights tensor has + * been reshaped with @ref CLWeightsReshapeKernel. * */ void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *weights, @@ -154,24 +157,26 @@ public: unsigned int invalid_right, unsigned int invalid_bottom, const WeightsInfo &weights_info = WeightsInfo()); /** Static function to check if given info will lead to a valid configuration of @ref - * CLDirectTransposeConvLayer + * CLDirectTransposeConvLayer * - * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an - * optional 4th dimension for batch of inputs. - * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. - * @param[in] weights The 4d weights info with dimensions [width, height, IFM, OFM]. Data - * type supported: Same as @p input. - * @param[in] bias (Optional) The biases have one dimension. - * Data type supported: Should match @p input data type, except for input - * of QASYMM8 and QASYMM8_SIGNED type where biases should be of S32 type - * @param[in] output Output tensor info. The output has the same number of dimensions as the - * @p input. - * @param[in] info Contains padding and policies to be used in the deconvolution, this is - * decribed in @ref PadStrideInfo. - * @param[in] invalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. - * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, - * specifies if the weights tensor has been reshaped with @ref CLWeightsReshapeKernel. + * @param[in] input Input tensor info. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. + * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. + * @param[in] weights The 4d weights info with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. + * @param[in] bias (Optional) The biases have one dimension. + * Data type supported: Should match @p input data type, + * except for input of QASYMM8 and QASYMM8_SIGNED type + * where biases should be of S32 type + * @param[in] output Output tensor info. The output has the same number of dimensions + * as the @p input. + * @param[in] info Contains padding and policies to be used in the deconvolution, + * this is decribed in @ref PadStrideInfo. + * @param[in] invalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, + * specifies if the weights tensor has been reshaped + * with @ref CLWeightsReshapeKernel. * * @return a status */ diff --git a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedLayerEx.h b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedLayerEx.h index e65a646dc..f27e9913e 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedLayerEx.h +++ b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedLayerEx.h @@ -216,7 +216,7 @@ private: CLConvertFullyConnectedWeights _convert_weights; weights_transformations::CLConvertFullyConnectedWeightsManaged _convert_weights_managed; weights_transformations::CLFullyConnectedLayerReshapeWeightsExManaged - _reshape_weights_managed_function; + _reshape_weights_managed_function; CLFlattenLayer _flatten_layer; CLFullyConnectedLayerReshapeWeightsEx _reshape_weights_function; CLGEMM _mm_gemm; diff --git a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedReshapingLayer.h b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedReshapingLayer.h index 289ab167f..bdb168664 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedReshapingLayer.h +++ b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLFullyConnectedReshapingLayer.h @@ -43,8 +43,8 @@ public: public: CLFullyConnectedReshapingLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr) - : _input(nullptr), _weights(nullptr), _biases(nullptr), _output(nullptr), _cl_buffer{}, - _memory_manager{memory_manager}, _cl_fc{nullptr}, _cl_reshape{}, _needs_reshape(false) + : _input(nullptr), _weights(nullptr), _biases(nullptr), _output(nullptr), _cl_buffer{}, + _memory_manager{memory_manager}, _cl_fc{nullptr}, _cl_reshape{}, _needs_reshape(false) { // DO NOTHING } diff --git a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLGatherEx.h b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLGatherEx.h index b01ec4255..167554c9e 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLGatherEx.h +++ b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLGatherEx.h @@ -66,7 +66,7 @@ public: * @param[out] output The output tensor, Data types supported: same as @p input. * @param[in] axis (Optional) The axis in @p input to gather @p indices from. Defaults to 0 * @return N/A - */ + */ void configure(const ICLTensor *input, const ICLTensor *indices, ICLTensor *output, int axis = 0); /** diff --git a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLTransposeConvLayer.h b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLTransposeConvLayer.h index 5fb102e47..5b27d362a 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLTransposeConvLayer.h +++ b/compute/ARMComputeEx/arm_compute/runtime/CL/functions/CLTransposeConvLayer.h @@ -63,20 +63,22 @@ public: /** Set the input, weights, biases and output tensors. * - * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and an - * optional 4th dimension for batch of inputs. Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. - * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data type - * supported: Same as @p input. - * @param[in] bias (Optional) The biases have one dimension. Data type supported: Same - * as @p input. - * @param[out] output Output tensor. The output has the same number of dimensions as the - * @p input. - * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, this - * is described in @ref PadStrideInfo. - * @param[in] invalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. - * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, - * specifies if the weights tensor has been reshaped with @ref CLWeightsReshapeKernel. + * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. + * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. + * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. + * @param[in] bias (Optional) The biases have one dimension. + * Data type supported: Same as @p input. + * @param[out] output Output tensor. The output has the same number of dimensions + * as the @p input. + * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, + * this is described in @ref PadStrideInfo. + * @param[in] invalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in] weights_info (Optional) Weights information needed for + * @ref CLConvolutionLayer, specifies if the weights tensor has + * been reshaped with @ref CLWeightsReshapeKernel. * */ void configure(ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, @@ -85,22 +87,22 @@ public: /** Set the input, weights, biases and output tensors. * * @param[in] compile_context The compile context to be used. - * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and - * an optional 4th dimension for batch of inputs. Data types supported: - * QASYMM8_SIGNED/QASYMM8/F16/F32. - * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data - * type supported: Same as @p input. - * @param[in] bias (Optional) The biases have one dimension. Data type supported: - * Same as @p input. - * @param[out] output Output tensor. The output has the same number of dimensions as - * the @p input. + * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. + * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. + * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. + * @param[in] bias (Optional) The biases have one dimension. + * Data type supported: Same as @p input. + * @param[out] output Output tensor. The output has the same number of dimensions + * as the @p input. * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, - * this is described in @ref PadStrideInfo. - * @param[in] invalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. - * @param[in] weights_info (Optional) Weights information needed for @ref - * CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref - * CLWeightsReshapeKernel. + * this is described in @ref PadStrideInfo. + * @param[in] invalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in] weights_info (Optional) Weights information needed for + * @ref CLConvolutionLayer, specifies if the weights tensor has + * been reshaped with @ref CLWeightsReshapeKernel. * */ void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *weights, @@ -108,22 +110,24 @@ public: unsigned int invalid_right, unsigned int invalid_bottom, const WeightsInfo &weights_info = WeightsInfo()); /** Static function to check if given info will lead to a valid configuration of @ref - * CLTransposeConvLayer + * CLTransposeConvLayer * - * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an - * optional 4th dimension for batch of inputs. Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. - * @param[in] weights The 4d weights info with dimensions [width, height, IFM, OFM]. Data - * type supported: Same as @p input. - * @param[in] bias (Optional) The biases have one dimension. Data type supported: Same as - * @p input. - * @param[in] output Output tensor info. The output has the same number of dimensions as the - * @p input. - * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, this is - * described in @ref PadStrideInfo. - * @param[in] invalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. - * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, - * specifies if the weights tensor has been reshaped with @ref CLWeightsReshapeKernel. + * @param[in] input Input tensor info. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. + * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. + * @param[in] weights The 4d weights info with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. + * @param[in] bias (Optional) The biases have one dimension. + * Data type supported: Same as @p input. + * @param[in] output Output tensor info. The output has the same number of dimensions + * as the @p input. + * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, + * this is described in @ref PadStrideInfo. + * @param[in] invalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, + * specifies if the weights tensor has been reshaped with + * @ref CLWeightsReshapeKernel. * * @return a status */ diff --git a/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEFullyConnectedReshapingLayer.h b/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEFullyConnectedReshapingLayer.h index 18cb61bf9..e34b4dcb0 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEFullyConnectedReshapingLayer.h +++ b/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEFullyConnectedReshapingLayer.h @@ -43,8 +43,8 @@ public: public: NEFullyConnectedReshapingLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr) - : _memory_manager{memory_manager}, _input(nullptr), _weights(nullptr), _biases(nullptr), - _output(nullptr), _neon_buffer{}, _neon_fc{nullptr}, _neon_reshape{}, _needs_reshape(false) + : _memory_manager{memory_manager}, _input(nullptr), _weights(nullptr), _biases(nullptr), + _output(nullptr), _neon_buffer{}, _neon_fc{nullptr}, _neon_reshape{}, _needs_reshape(false) { // DO NOTHING } diff --git a/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEOneHot.h b/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEOneHot.h index b2ea6270f..1a68f801a 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEOneHot.h +++ b/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEOneHot.h @@ -66,19 +66,20 @@ public: void configure(const ITensor *indices, const ITensor *depth, const ITensor *on_value, const ITensor *off_value, ITensor *output, int axis = -1); /** Static function to check if given info will lead to a valid configuration of @ref - * NEOneHotKernel + * NEOneHotKernel * - * @param[in] indices Indices tensor info. Supported tensor rank: up to 3. Must be one of the - * following types: U32/S32 - * @param[in] depth The tensor info for depth of the one hot dimension. Supported tensor rank: - * up to 3. Must be one of the following types: U32/S32 - * @param[in] on_value On value tensor info. Supported tensor rank: only 1. Data type supported: - * U8/S8/U16/S16/F16/U32/S32/F32 - * @param[in] off_value Off value tensor info. Supported tensor rank: only 1. Data type supported: - * Same as @p on_value - * @param[out] output Destination tensor info. Data type supported: Same as @p on_value - * @param[in] axis (Optional) The axis to fill. Negative values wrap around. Defaults to -1. - * The value must be in range [-indices.rank , indices.rank) + * @param[in] indices Indices tensor info. Supported tensor rank: up to 3. + * Must be one of the following types: U32/S32 + * @param[in] depth The tensor info for depth of the one hot dimension. + * Supported tensor rank: up to 3. + * Must be one of the following types: U32/S32 + * @param[in] on_value On value tensor info. Supported tensor rank: only 1. + * Data type supported: U8/S8/U16/S16/F16/U32/S32/F32 + * @param[in] off_value Off value tensor info. Supported tensor rank: only 1. + * Data type supported: Same as @p on_value + * @param[out] output Destination tensor info. Data type supported: Same as @p on_value + * @param[in] axis (Optional) The axis to fill. Negative values wrap around. Defaults to -1. + * The value must be in range [-indices.rank , indices.rank) * * @return a status */ diff --git a/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NETransposeConvLayer.h b/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NETransposeConvLayer.h index 24ff5dac9..7a08dae97 100644 --- a/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NETransposeConvLayer.h +++ b/compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NETransposeConvLayer.h @@ -110,39 +110,42 @@ public: /** Set the input, weights, biases and output tensors. * - * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and an - * optional 4th dimension for batch of inputs. Data types supported: F32/F16/QASYMM8/QASYMM8_SIGNED. - * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data type - * supported: Same as @p input. - * @param[in] bias Optional, ignored if NULL. The biases have one dimension. Data type - * supported: Data types supported: S32 for QASYMM8 and QASYMM8_SIGNED input, F32 for F32 input, F16 - * for F16 input. - * @param[out] output Output tensor. The output has the same number of dimensions as the @p - * input. - * @param[in] info Contains padding and policies to be used in the deconvolution, this is - * decribed in @ref PadStrideInfo. - * @param[in] invalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. + * Data types supported: F32/F16/QASYMM8/QASYMM8_SIGNED. + * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. + * @param[in] bias Optional, ignored if NULL. The biases have one dimension. + * Data type supported: Data types supported: S32 for QASYMM8 and + * QASYMM8_SIGNED input, F32 for F32 input, F16 for F16 input. + * @param[out] output Output tensor. The output has the same number of dimensions as + * the @p input. + * @param[in] info Contains padding and policies to be used in the deconvolution, + * this is decribed in @ref PadStrideInfo. + * @param[in] invalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. * */ void configure(ITensor *input, const ITensor *weights, const ITensor *bias, ITensor *output, const PadStrideInfo &info, unsigned int invalid_right, unsigned int invalid_bottom); /** Static function to check if given info will lead to a valid configuration of @ref - * NETransposeConvLayer + * NETransposeConvLayer * - * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an - * optional 4th dimension for batch of inputs. Data types supported: F32/F16/QASYMM8/QASYMM8_SIGNED. - * @param[in] weights The 4d weights info with dimensions [width, height, IFM, OFM]. Data type - * supported: Same as @p input. - * @param[in] bias (Optional) The biases have one dimension. Data type supported: Data types - * supported: S32 for QASYMM8 and QASYMM8_SIGNED input, F32 for F32 input, F16 for F16 input. - * @param[in] output Output tensor info. The output has the same number of dimensions as the @p - * input. - * @param[in] info Contains padding and policies to be used in the deconvolution, this is - * decribed in @ref PadStrideInfo. - * @param[in] innvalid_right The number of zeros added to right edge of the output. - * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. + * @param[in] input Input tensor info. 3 lower dimensions represent a single input, + * and an optional 4th dimension for batch of inputs. + * Data types supported: F32/F16/QASYMM8/QASYMM8_SIGNED. + * @param[in] weights The 4d weights info with dimensions [width, height, IFM, OFM]. + * Data type supported: Same as @p input. + * @param[in] bias (Optional) The biases have one dimension. + * Data types supported: S32 for QASYMM8 and QASYMM8_SIGNED input, + * F32 for F32 input, F16 for F16 input. + * @param[in] output Output tensor info. The output has the same number of dimensions as + * the @p input. + * @param[in] info Contains padding and policies to be used in the deconvolution, + * this is decribed in @ref PadStrideInfo. + * @param[in] innvalid_right The number of zeros added to right edge of the output. + * @param[in] invalid_bottom The number of zeros added to bottom edge of the output. * * @return a status */ |