diff options
Diffstat (limited to 'arm_compute/core/CL/kernels')
27 files changed, 78 insertions, 82 deletions
diff --git a/arm_compute/core/CL/kernels/CLActivationLayerKernel.h b/arm_compute/core/CL/kernels/CLActivationLayerKernel.h index c6d8f96a87..12d00de7e8 100644 --- a/arm_compute/core/CL/kernels/CLActivationLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLActivationLayerKernel.h @@ -51,7 +51,7 @@ public: * @note If the output tensor is a nullptr, the activation function will be performed in-place * * @param[in, out] input Source tensor. In case of @p output tensor = nullptr, this tensor will store the result - * of the activation function. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * of the activation function. Data types supported: QASYMM8/F16/F32. * @param[out] output Destination tensor. Data type supported: same as @p input * @param[in] act_info Activation layer information. */ @@ -59,7 +59,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLActivationLayerKernel * * @param[in] input Source tensor info. In case of @p output tensor info = nullptr, this tensor will store the result - * of the activation function. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * of the activation function. Data types supported: QASYMM8/F16/F32. * @param[in] output Destination tensor info. Data type supported: same as @p input * @param[in] act_info Activation layer information. * diff --git a/arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h b/arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h index a33cbf321f..f4275f4153 100644 --- a/arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h +++ b/arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h @@ -53,17 +53,17 @@ public: ~CLArithmeticAdditionKernel() = default; /** Initialise the kernel's inputs, output and convertion policy. * - * @param[in] input1 First tensor input. Data types supported: U8/QS8/QASYMM8/QS16/S16/F16/F32. - * @param[in] input2 Second tensor input. Data types supported: U8/QS8 (only if @p input1 is QS8), QASYMM8 (only if @p input1 is QASYMM8), QS16 (only if @p input1 is QS16), S16/F16/F32. - * @param[out] output Output tensor. Data types supported: U8 (Only if both inputs are U8), QS8 (only if both inputs are QS8),QASYMM8 (only if @p input1 is QASYMM8), QS16 (only if both inputs are QS16), S16/F16/F32. + * @param[in] input1 First tensor input. Data types supported: U8/QASYMM8/S16/F16/F32. + * @param[in] input2 Second tensor input. Data types supported: U8, QASYMM8 (only if @p input1 is QASYMM8), S16/F16/F32. + * @param[out] output Output tensor. Data types supported: U8 (Only if both inputs are U8), QASYMM8 (only if @p input1 is QASYMM8), S16/F16/F32. * @param[in] policy Policy to use to handle overflow. */ void configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy); /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticAdditionKernel * - * @param[in] input1 First tensor input info. Data types supported: U8/QS8/QASYMM8/QS16/S16/F16/F32. - * @param[in] input2 Second tensor input info. Data types supported: U8/QS8 (only if @p input1 is QS8), QASYMM8 (only if @p input1 is QASYMM8), QS16 (only if @p input1 is QS16), S16/F16/F32. - * @param[in] output Output tensor info. Data types supported: U8 (Only if both inputs are U8), QS8 (only if both inputs are QS8), QASYMM8 (only if both inputs are QASYMM8), QS16 (only if both inputs are QS16), S16/F16/F32. + * @param[in] input1 First tensor input info. Data types supported: U8/QASYMM8/S16/F16/F32. + * @param[in] input2 Second tensor input info. Data types supported: U8, QASYMM8 (only if @p input1 is QASYMM8), S16/F16/F32. + * @param[in] output Output tensor info. Data types supported: U8 (Only if both inputs are U8), QASYMM8 (only if both inputs are QASYMM8), S16/F16/F32. * @param[in] policy Policy to use to handle overflow. * * @return a status diff --git a/arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h b/arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h index c5f862a61f..35b918fe4b 100644 --- a/arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h +++ b/arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -55,17 +55,17 @@ public: /** Initialise the kernel's inputs, output and convertion policy. * - * @param[in] input1 First tensor input. Data types supported: U8/QS8/QS16/S16/F16/F32. - * @param[in] input2 Second tensor input. Data types supported: U8/QS8 (only if @p input1 is QS8), QS16 (only if @p input1 is QS16), S16/F16/F32. - * @param[out] output Output tensor. Data types supported: U8 (Only if both inputs are U8), QS8 (only if both inputs are QS8), QS16 (only if both inputs are QS16), S16/F16/F32. + * @param[in] input1 First tensor input. Data types supported: U8/S16/F16/F32. + * @param[in] input2 Second tensor input. Data types supported: U8/S16/F16/F32. + * @param[out] output Output tensor. Data types supported: U8 (Only if both inputs are U8), S16/F16/F32. * @param[in] policy Policy to use to handle overflow. */ void configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy); /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticSubtractionKernel * - * @param[in] input1 First tensor input info. Data types supported: U8/QS8/QS16/S16/F16/F32. - * @param[in] input2 Second tensor input info. Data types supported: U8/QS8 (only if @p input1 is QS8), QS16 (only if @p input1 is QS16), S16/F16/F32. - * @param[in] output Output tensor info. Data types supported: U8 (Only if both inputs are U8), QS8 (only if both inputs are QS8), QS16 (only if both inputs are QS16), S16/F16/F32. + * @param[in] input1 First tensor input info. Data types supported: U8/S16/F16/F32. + * @param[in] input2 Second tensor input info. Data types supported: U8/S16/F16/F32. + * @param[in] output Output tensor info. Data types supported: U8 (Only if both inputs are U8), S16/F16/F32. * @param[in] policy Policy to use to handle overflow. * * @return a status diff --git a/arm_compute/core/CL/kernels/CLBatchNormalizationLayerKernel.h b/arm_compute/core/CL/kernels/CLBatchNormalizationLayerKernel.h index 8015f08d1b..9c8d02532a 100644 --- a/arm_compute/core/CL/kernels/CLBatchNormalizationLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLBatchNormalizationLayerKernel.h @@ -54,7 +54,7 @@ public: * * @param[in, out] input Source tensor. In case of @p output tensor = nullptr, this tensor will store the result. * 3 lower dimensions represent a single input with dimensions [width, height, FM]. - * The rest are optional and used for representing batches. Data types supported: QS8/QS16/F16/F32. + * The rest are optional and used for representing batches. Data types supported: F16/F32. * @param[out] output Destination tensor. Output will have the same number of dimensions as input. Data type supported: same as @p input * @param[in] mean Mean values tensor. 1 dimension with size equal to the feature maps [FM]. Data types supported: Same as @p input * @param[in] var Variance values tensor. 1 dimension with size equal to the feature maps [FM]. Data types supported: Same as @p input @@ -69,7 +69,7 @@ public: * * @param[in] input Source tensor info. In case of @p output tensor info = nullptr, this tensor will store the result. * 3 lower dimensions represent a single input with dimensions [width, height, FM]. - * The rest are optional and used for representing batches. Data types supported: QS8/QS16/F16/F32. + * The rest are optional and used for representing batches. Data types supported: F16/F32. * @param[in] output Destination tensor info. Output will have the same number of dimensions as input. Data type supported: same as @p input * @param[in] mean Mean values tensor info. 1 dimension with size equal to the feature maps [FM]. Data types supported: Same as @p input * @param[in] var Variance values tensor info. 1 dimension with size equal to the feature maps [FM]. Data types supported: Same as @p input diff --git a/arm_compute/core/CL/kernels/CLChannelShuffleLayerKernel.h b/arm_compute/core/CL/kernels/CLChannelShuffleLayerKernel.h index 684a0e5027..f7bd205ec7 100644 --- a/arm_compute/core/CL/kernels/CLChannelShuffleLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLChannelShuffleLayerKernel.h @@ -48,14 +48,14 @@ public: ~CLChannelShuffleLayerKernel() = default; /** Configure function's inputs and outputs. * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[out] output Output tensor. Data type supported: Same as @p input * @param[in] num_groups Number of groups. Must be greater than 1 and the number of channels of the tensors must be a multiple of the number of groups. */ void configure(const ICLTensor *input, ICLTensor *output, unsigned int num_groups); /** Static function to check if given info will lead to a valid configuration of @ref CLChannelShuffleLayerKernel * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[out] output Output tensor. Data type supported: Same as @p input * @param[in] num_groups Number of groups. Must be greater than 1 and the number of channels of the tensors must be a multiple of the number of groups. * diff --git a/arm_compute/core/CL/kernels/CLCol2ImKernel.h b/arm_compute/core/CL/kernels/CLCol2ImKernel.h index 3779325efe..94f21b1ebc 100644 --- a/arm_compute/core/CL/kernels/CLCol2ImKernel.h +++ b/arm_compute/core/CL/kernels/CLCol2ImKernel.h @@ -66,7 +66,7 @@ public: /** Set the input and output of the kernel. * - * @param[in] input The input tensor to convert. Data types supported: QS8/QS16/QASYMM8/F16/F32 + * @param[in] input The input tensor to convert. Data types supported: QASYMM8/F16/F32 * @param[out] output The output tensor. 3 lower dimensions represent a single output [width, height, OFM], * while the rest represent batch of outputs. Data types supported: Same as @p input * @param[in] convolved_dims Output convolved dimensions. @@ -74,7 +74,7 @@ public: void configure(const ICLTensor *input, ICLTensor *output, std::pair<unsigned int, unsigned int> convolved_dims); /** Static function to check if given info will lead to a valid configuration of @ref CLCol2ImKernel * - * @param[in] input The input tensor to convert. Data types supported: QS8/QS16/QASYMM8/F16/F32 + * @param[in] input The input tensor to convert. Data types supported: QASYMM8/F16/F32 * @param[in] output The output tensor. 3 lower dimensions represent a single output [width, height, OFM], * while the rest represent batch of outputs. Data types supported: Same as @p input * @param[in] convolved_dims Output convolved dimensions. diff --git a/arm_compute/core/CL/kernels/CLConvertFullyConnectedWeightsKernel.h b/arm_compute/core/CL/kernels/CLConvertFullyConnectedWeightsKernel.h index fe24aa9d8c..f5e2f0de89 100644 --- a/arm_compute/core/CL/kernels/CLConvertFullyConnectedWeightsKernel.h +++ b/arm_compute/core/CL/kernels/CLConvertFullyConnectedWeightsKernel.h @@ -55,7 +55,7 @@ public: ~CLConvertFullyConnectedWeightsKernel() = default; /** Set the input and output tensor. * - * @param[in] input Source weights tensor to convert. Must be 2 dimensional. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/U32/S32/QS32/F16/F32. + * @param[in] input Source weights tensor to convert. Must be 2 dimensional. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/QS32/F16/F32. * @param[out] output The converted weights tensor. Shape and Data Type: Same as @p input. * @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer). Must be in NCHW format. * @param[in] data_layout The data layout the weights have been trained in. @@ -63,7 +63,7 @@ public: void configure(const ICLTensor *input, ICLTensor *output, const TensorShape &original_input_shape, DataLayout data_layout); /** Static function to check if given info will lead to a valid configuration of @ref CLConvertFullyConnectedWeightsKernel * - * @param[in] input Source weights tensor info to convert. Must be 2 dimensional. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/U32/S32/QS32/F16/F32. + * @param[in] input Source weights tensor info to convert. Must be 2 dimensional. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/QS32/F16/F32. * @param[in] output The converted weights tensor info. Shape and Data Type: Same as @p input. * @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer). Must be in NCHW format. * @param[in] data_layout The data layout the weights have been trained in. diff --git a/arm_compute/core/CL/kernels/CLDepthConcatenateLayerKernel.h b/arm_compute/core/CL/kernels/CLDepthConcatenateLayerKernel.h index 467bdfab3b..cbcab8f554 100644 --- a/arm_compute/core/CL/kernels/CLDepthConcatenateLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLDepthConcatenateLayerKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -52,7 +52,7 @@ public: ~CLDepthConcatenateLayerKernel() = default; /** Initialise the kernel's inputs and output * - * @param[in] input Input tensor. Data types supported: QS8/QS16/F16/F32. + * @param[in] input Input tensor. Data types supported: F16/F32. * @param[in] depth_offset The offset on the Z axis. * @param[in,out] output Output tensor. Data types supported: Same as @p input. * diff --git a/arm_compute/core/CL/kernels/CLDepthConvertLayerKernel.h b/arm_compute/core/CL/kernels/CLDepthConvertLayerKernel.h index 3a6310d69e..7e795c607a 100644 --- a/arm_compute/core/CL/kernels/CLDepthConvertLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLDepthConvertLayerKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -43,17 +43,14 @@ public: * * Valid conversions Input -> Output : * - * - QS8 -> F32 - * - QS16 -> F32 * - U8 -> U16, S16, U32, S32 * - U16 -> U8, U32, S32 * - S16 -> U8, U32, S32 * - U32 -> U8, U16, S16 * - S32 -> U8, U16, S16 - * - F32 -> QS8, QS16 * - * @param[in] input The input tensor to convert. Data types supported: U8/QS8/U16/S16/QS16/U32/S32/F32. - * @param[out] output The output tensor. Data types supported: U8/QS8/U16/S16/QS16/U32/S32/F32. + * @param[in] input The input tensor to convert. Data types supported: U8/U16/S16/U32/S32/F32. + * @param[out] output The output tensor. Data types supported: U8/U16/S16/U32/S32/F32. * @param[in] policy Conversion policy * @param[in] shift Value for down/up conversions. Must be 0 <= shift < 8. */ diff --git a/arm_compute/core/CL/kernels/CLDirectConvolutionLayerKernel.h b/arm_compute/core/CL/kernels/CLDirectConvolutionLayerKernel.h index eb1bf58b1b..bd37e35334 100644 --- a/arm_compute/core/CL/kernels/CLDirectConvolutionLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLDirectConvolutionLayerKernel.h @@ -56,7 +56,7 @@ public: * 5x5 convolution with stride_x = 1/2, stride_y = 1/2 * * @param[in] input The input tensor to convolve. 3 lower dimensions represent a single input [width, height, IFM], - * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/QS8/QS16/F16/F32. + * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. * The 3rd dimension must be the same as the input's volume 3rd dimension. * Data type supported:Same as @p input. @@ -70,7 +70,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLDirectConvolutionLayerKernel * * @param[in] input The input tensor to convolve. 3 lower dimensions represent a single input [width, height, IFM], - * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. * The 3rd dimension must be the same as the input's volume 3rd dimension. * Data type supported:Same as @p input. diff --git a/arm_compute/core/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h b/arm_compute/core/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h index 9340e9a8d8..1947a98ba3 100644 --- a/arm_compute/core/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h +++ b/arm_compute/core/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h @@ -51,11 +51,11 @@ public: /** Set the accumulate buffer and the biases of the kernel. * * @param[in, out] input Input to add the bias to. If @p output is not specified then accumulation is done in-place. - * Data type supported: S32/QS16/QS32/F16/F32 + * Data type supported: S32/QS32/F16/F32 * @param[in] bias (Optional) The shared bias tensor to add. It must be 1D Tensor. Data type supported: Same as @p input * @param[out] output (Optional) If the output tensor is specified the accumulation is done out-of-place. (Defaults to nullptr) * Required parameter if output is of QASYMM8 type. - * Data types supported: QS8/QASYMM8/QS16/F16/F32 + * Data types supported: QASYMM8/F16/F32 * @param[in] result_fixedpoint_multiplier (Optional)Fixed point value to be multiplied to each element of the input matrix when once the result_offset has been add * @param[in] result_shift (Optional)Integer value used to round to nearest division by a power-of-two the result after the fixed point multiplication * @param[in] result_offset_after_shift (Optional)Offset to be applied to result before converting it back to QASYMM8 @@ -65,10 +65,10 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLDirectConvolutionLayerOutputStageKernel * * @param[in] input Input to add the bias to. If @p output is not specified then accumulation is done in-place. - * Data type supported: QS16/QS32/F16/F32 + * Data type supported: QS32/F16/F32 * @param[in] bias (Optional) The shared bias tensor to add. It must be 1D Tensor. Data type supported: Same as @p input * @param[in] output (Optional) If the output tensor is specified the accumulation is done out-of-place. (Defaults to nullptr) - * Data type supported: QS8/QS16/F16/F32 + * Data type supported: F16/F32 * @return a status */ static Status validate(const ITensorInfo *input, const ITensorInfo *bias = nullptr, const ITensorInfo *output = nullptr); diff --git a/arm_compute/core/CL/kernels/CLFillBorderKernel.h b/arm_compute/core/CL/kernels/CLFillBorderKernel.h index 18031c7e7e..20e872eccb 100644 --- a/arm_compute/core/CL/kernels/CLFillBorderKernel.h +++ b/arm_compute/core/CL/kernels/CLFillBorderKernel.h @@ -51,7 +51,7 @@ public: /** Initialise the kernel's input, output and border mode. * - * @param[in,out] tensor Tensor to process Data types supported: U8/QS8/S16/QS16/S32/F16/F32. + * @param[in,out] tensor Tensor to process Data types supported: U8/S16/S32/F16/F32. * @param[in] border_size Size of the border to fill in elements. * @param[in] border_mode Border mode to use for the convolution. * @param[in] constant_border_value (Optional) Constant value to use for borders if border_mode is set to CONSTANT. diff --git a/arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h b/arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h index c0fef45afe..7f8e766f1a 100644 --- a/arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h +++ b/arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h @@ -64,14 +64,14 @@ public: CLGEMMInterleave4x4Kernel &operator=(CLGEMMInterleave4x4Kernel &&) = default; /** Initialise the kernel's input and output. * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[out] output Output tensor. Data type supported: same as @p input * @param[in] mult_interleave4x4_height (Optional) Multiplication factor for the height of the 4x4 interleave block */ void configure(const ICLTensor *input, ICLTensor *output, int mult_interleave4x4_height = 1); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMInterleave4x4Kernel * - * @param[in] input Input tensor info. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] output Output tensor info which stores the interleaved matrix. Data type supported: same as @p input. * @param[in] mult_interleave4x4_height Multiplication factor for the height of the 4x4 interleave block * diff --git a/arm_compute/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.h b/arm_compute/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.h index 2956f93cdc..f201af0d5e 100644 --- a/arm_compute/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.h +++ b/arm_compute/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -46,13 +46,13 @@ public: CLGEMMMatrixAccumulateBiasesKernel &operator=(CLGEMMMatrixAccumulateBiasesKernel &&) = default; /** Set the accumulate buffer and the biases of the kernel. * - * @param[in, out] accum The accumulate tensor to convert. Data types supported: QS8/QS16/F16/F32 + * @param[in, out] accum The accumulate tensor to convert. Data types supported: F16/F32 * @param[in] biases The shared biases tensor to append. It must be 1D tensor. Data types supported: Same as @p input */ void configure(ICLTensor *accum, const ICLTensor *biases); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMMatrixAccumulateBiasesKernel * - * @param[in] accum The accumulate tensor to convert. Data types supported: QS8/QS16/F16/F32 + * @param[in] accum The accumulate tensor to convert. Data types supported: F16/F32 * @param[in] biases The shared biases tensor to append. It must be 1D tensor. Data types supported: Same as @p input * @param[in] gpu_target GPU target * diff --git a/arm_compute/core/CL/kernels/CLGEMMMatrixAdditionKernel.h b/arm_compute/core/CL/kernels/CLGEMMMatrixAdditionKernel.h index 3755d943c5..bf8e1d4b17 100644 --- a/arm_compute/core/CL/kernels/CLGEMMMatrixAdditionKernel.h +++ b/arm_compute/core/CL/kernels/CLGEMMMatrixAdditionKernel.h @@ -52,14 +52,14 @@ public: * * @note The input and output tensors must have the same dimensions * - * @param[in] input Input tensor (Matrix C). Data types supported: QS8/QS16/F16/F32 + * @param[in] input Input tensor (Matrix C). Data types supported: F16/F32 * @param[in, out] output Output tensor. If this kernel is used to finalize the GEMM result (alpha * AB + beta * C), output must contain the result obtained by @ref CLGEMMMatrixMultiplyKernel. Data type supported: same as @p input * @param[in] beta Weight of matrix C */ void configure(const ICLTensor *input, ICLTensor *output, float beta); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMMatrixAdditionKernel. * - * @param[in] input Input tensor (Matrix C). Data types supported: QS8/QS16/F16/F32 + * @param[in] input Input tensor (Matrix C). Data types supported: F16/F32 * @param[in] output Output tensor. If this kernel is used to finalize the GEMM result (alpha * AB + beta * C), output must contain the result obtained by @ref CLGEMMMatrixMultiplyKernel. Data type supported: same as @p input * @param[in] beta Weight of matrix C * diff --git a/arm_compute/core/CL/kernels/CLGEMMMatrixMultiplyKernel.h b/arm_compute/core/CL/kernels/CLGEMMMatrixMultiplyKernel.h index 15bba0cd0f..1b6a0c87a9 100644 --- a/arm_compute/core/CL/kernels/CLGEMMMatrixMultiplyKernel.h +++ b/arm_compute/core/CL/kernels/CLGEMMMatrixMultiplyKernel.h @@ -53,7 +53,7 @@ public: CLGEMMMatrixMultiplyKernel &operator=(CLGEMMMatrixMultiplyKernel &&) = default; /** Initialise the kernel's input, output and alpha * - * @param[in] input0 Input tensor containing the Matrix A. Data types supported: QS8/QS16/F16/F32 + * @param[in] input0 Input tensor containing the Matrix A. Data types supported: F16/F32 * @param[in] input1 Input tensor containing the Matrix B. Data type supported: same as @p input0 * @param[out] output Output tensor to store the result of matrix multiplication. Data type supported: same as @p input0 * @param[in] alpha Weight of the matrix product @@ -64,7 +64,7 @@ public: void configure(const ICLTensor *input0, const ICLTensor *input1, ICLTensor *output, float alpha, bool is_interleaved_transposed = true, const GEMMReshapeInfo &reshape_info = GEMMReshapeInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMMatrixMultiplyKernel * - * @param[in] input0 Input tensor containing the Matrix A. Data types supported: QS8/QS16/F16/F32 + * @param[in] input0 Input tensor containing the Matrix A. Data types supported: F16/F32 * @param[in] input1 Input tensor containing the Matrix B. Data type supported: same as @p input0 * @param[in] output Output tensor to store the result of matrix multiplication. Data type supported: same as @p input0 * @param[in] alpha Weight of the matrix product diff --git a/arm_compute/core/CL/kernels/CLGEMMTranspose1xWKernel.h b/arm_compute/core/CL/kernels/CLGEMMTranspose1xWKernel.h index 9a3069eab6..47a4ad515b 100644 --- a/arm_compute/core/CL/kernels/CLGEMMTranspose1xWKernel.h +++ b/arm_compute/core/CL/kernels/CLGEMMTranspose1xWKernel.h @@ -70,14 +70,14 @@ class CLGEMMTranspose1xWKernel : public ICLSimple2DKernel public: /** Initialise the kernel's input and output. * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[out] output Output tensor. Data type supported: same as @p input * @param[in] mult_transpose1xW_width (Optional) Multiplication factor for the width of the 1xW transposed block */ void configure(const ICLTensor *input, ICLTensor *output, int mult_transpose1xW_width = 1); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMTranspose1xWKernel * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] output Output tensor. Data type supported: same as @p input. * @param[in] mult_transpose1xW_width Multiplication factor for the width of the 1xW transposed block * diff --git a/arm_compute/core/CL/kernels/CLIm2ColKernel.h b/arm_compute/core/CL/kernels/CLIm2ColKernel.h index 7e119a32a8..fc930abcbe 100644 --- a/arm_compute/core/CL/kernels/CLIm2ColKernel.h +++ b/arm_compute/core/CL/kernels/CLIm2ColKernel.h @@ -69,7 +69,7 @@ public: /** Set the input and output of the kernel. * * @param[in] input The input tensor to convert. 3 lower dimensions represent a single input [width, height, IFM], - * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QS8/QASYMM8/QS16/F16/F32 + * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F16/F32 * @param[out] output The output tensor. First 2 lower dimensions represent a transform of each 3D input, * while every dimension above represents a batch. Data types supported: Same as @p input * @param[in] kernel_dims The kernel dimensions (width and height). @@ -81,7 +81,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLIm2ColKernel * * @param[in] input The input tensor to convert. 3 lower dimensions represent a single input [width, height, IFM], - * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QS8/QASYMM8/QS16/F16/F32 + * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F16/F32 * @param[in] output The output tensor. First 2 lower dimensions represent a transform of each 3D input, * while every dimension above represents a batch. Data types supported: Same as @p input * @param[in] kernel_dims The kernel dimensions (width and height). @@ -113,7 +113,7 @@ private: /** Chooses and configure the right kernel for the given input arguments. * * @param[in] input The input tensor to convert. 3 lower dimensions represent a single input [width, height, IFM], - * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QS8/QASYMM8/QS16/F16/F32 + * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F16/F32 * @param[in] output The output tensor. First 2 lower dimensions represent a transform of each 3D input, * while every dimension above represents a batch. Data types supported: Same as @p input * @param[in] kernel_dims The kernel dimensions (width and height). diff --git a/arm_compute/core/CL/kernels/CLNormalizationLayerKernel.h b/arm_compute/core/CL/kernels/CLNormalizationLayerKernel.h index ef00e59e5c..f2d37a781c 100644 --- a/arm_compute/core/CL/kernels/CLNormalizationLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLNormalizationLayerKernel.h @@ -48,7 +48,7 @@ public: /** Set the input and output tensors. * * @param[in] input Source tensor. 3 lower dims represent a single input with dimensions [width, height, IFM], - * and an optional 4th dimension for batch of inputs. Data types supported: QS8/QS16/F16/F32. + * and an optional 4th dimension for batch of inputs. Data types supported: F16/F32. * @param[out] output Destination tensor. Output will have the same number of dimensions as input. Data types supported: same as @p input. * @param[in] norm_info Normalization layer information like the normalization type, normalization size and other parameters. */ @@ -56,7 +56,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLNormalizationLayerKernel * * @param[in] input Source tensor. 3 lower dims represent a single input with dimensions [width, height, IFM], - * and an optional 4th dimension for batch of inputs. Data types supported: QS8/QS16/F16/F32. + * and an optional 4th dimension for batch of inputs. Data types supported: F16/F32. * @param[in] output Destination tensor. Output will have the same number of dimensions as input. Data types supported: same as @p input. * @param[in] norm_info Normalization layer information like the normalization type, normalization size and other parameters. * diff --git a/arm_compute/core/CL/kernels/CLPermuteKernel.h b/arm_compute/core/CL/kernels/CLPermuteKernel.h index b01df64ebd..21da141c0d 100644 --- a/arm_compute/core/CL/kernels/CLPermuteKernel.h +++ b/arm_compute/core/CL/kernels/CLPermuteKernel.h @@ -49,14 +49,14 @@ public: CLPermuteKernel &operator=(CLPermuteKernel &&) = default; /** Set the input and output of the kernel. * - * @param[in] input The input tensor to permute. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input The input tensor to permute. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] output The output tensor. Data types supported: Same as @p input * @param[in] perm Permutation vector */ void configure(const ICLTensor *input, ICLTensor *output, const PermutationVector &perm); /** Static function to check if given info will lead to a valid configuration of @ref CLPermuteKernel * - * @param[in] input First tensor input info. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32. + * @param[in] input First tensor input info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32. * @param[in] output Output tensor info. Data types supported: same as @p input. * @param[in] perm Permutation vector * diff --git a/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h b/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h index fcabb614df..b835aa701b 100644 --- a/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h +++ b/arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h @@ -49,11 +49,11 @@ public: CLPixelWiseMultiplicationKernel &operator=(CLPixelWiseMultiplicationKernel &&) = default; /** Initialise the kernel's input, output and border mode. * - * @param[in] input1 An input tensor. Data types supported: U8/QS8/QS16/S16/F16/F32. + * @param[in] input1 An input tensor. Data types supported: U8/S16/F16/F32. * @param[in] input2 An input tensor. Data types supported: same as @p input1. - * @param[out] output The output tensor, Data types supported: same as @p input1. Note: U8 (QS8, QS16) requires both inputs to be U8 (QS8, QS16). + * @param[out] output The output tensor, Data types supported: same as @p input1. Note: U8 requires both inputs to be U8. * @param[in] scale Scale to apply after multiplication. - * Scale must be positive and its value must be either 1/255 or 1/2^n where n is between 0 and 15. For QS8 and QS16 scale must be 1. + * Scale must be positive and its value must be either 1/255 or 1/2^n where n is between 0 and 15. * @param[in] overflow_policy Overflow policy. Supported overflow policies: Wrap, Saturate * @param[in] rounding_policy Rounding policy. Supported rounding modes: to zero, to nearest even. */ @@ -61,11 +61,11 @@ public: ConvertPolicy overflow_policy, RoundingPolicy rounding_policy); /** Static function to check if given info will lead to a valid configuration of @ref CLPixelWiseMultiplicationKernel * - * @param[in] input1 An input tensor info. Data types supported: U8/QS8/QS16/S16/F16/F32. + * @param[in] input1 An input tensor info. Data types supported: U8/S16/F16/F32. * @param[in] input2 An input tensor info. Data types supported: same as @p input1. - * @param[in] output The output tensor info, Data types supported: same as @p input1. Note: U8 (QS8, QS16) requires both inputs to be U8 (QS8, QS16). + * @param[in] output The output tensor info, Data types supported: same as @p input1. Note: U8 requires both inputs to be U8. * @param[in] scale Scale to apply after multiplication. - * Scale must be positive and its value must be either 1/255 or 1/2^n where n is between 0 and 15. For QS8 and QS16 scale must be 1. + * Scale must be positive and its value must be either 1/255 or 1/2^n where n is between 0 and 15. * @param[in] overflow_policy Overflow policy. Supported overflow policies: Wrap, Saturate * @param[in] rounding_policy Rounding policy. Supported rounding modes: to zero, to nearest even. * diff --git a/arm_compute/core/CL/kernels/CLPoolingLayerKernel.h b/arm_compute/core/CL/kernels/CLPoolingLayerKernel.h index c13507785b..db1a756229 100644 --- a/arm_compute/core/CL/kernels/CLPoolingLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLPoolingLayerKernel.h @@ -51,16 +51,15 @@ public: /** Set the input and output tensors. * - * @note QS8 and QS16 are supported only for pool sizes 3, 5 and 7 * - * @param[in] input Source tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Source tensor. Data types supported: QASYMM8/F16/F32. * @param[out] output Destination tensor. Data types supported: Same as @p input. * @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo. */ void configure(const ICLTensor *input, ICLTensor *output, const PoolingLayerInfo &pool_info); /** Static function to check if given info will lead to a valid configuration of @ref CLPoolingLayerKernel * - * @param[in] input Source tensor info. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Source tensor info. Data types supported: QASYMM8/F16/F32. * @param[in] output Destination tensor info. Data types supported: Same as @p input. * @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo. * diff --git a/arm_compute/core/CL/kernels/CLReshapeLayerKernel.h b/arm_compute/core/CL/kernels/CLReshapeLayerKernel.h index 044b5e7006..b253d66f4f 100644 --- a/arm_compute/core/CL/kernels/CLReshapeLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLReshapeLayerKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -49,7 +49,7 @@ public: ~CLReshapeLayerKernel() = default; /** Set the input and output of the kernel * - * @param[in] input Source tensor. Data type supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/U32/S32/F16/F32 + * @param[in] input Source tensor. Data type supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32 * @param[out] output Destination tensor. Data type supported: Same as @p input */ void configure(const ICLTensor *input, ICLTensor *output); diff --git a/arm_compute/core/CL/kernels/CLSoftmaxLayerKernel.h b/arm_compute/core/CL/kernels/CLSoftmaxLayerKernel.h index c562565175..b272878fe7 100644 --- a/arm_compute/core/CL/kernels/CLSoftmaxLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLSoftmaxLayerKernel.h @@ -38,13 +38,13 @@ class CLLogits1DMaxKernel : public ICLSimple3DKernel public: /** Set the input and output tensors. * - * @param[in] input Source tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32 + * @param[in] input Source tensor. Data types supported: QASYMM8/F16/F32 * @param[out] output Destination tensor. Data types supported: same as @p input */ void configure(const ICLTensor *input, ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLLogits1DMaxKernel * - * @param[in] input Source tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32 + * @param[in] input Source tensor. Data types supported: QASYMM8/F16/F32 * @param[in] output Destination tensor. Data types supported: same as @p input * * @return a status @@ -68,7 +68,7 @@ public: CLLogits1DShiftExpSumKernel &operator=(CLLogits1DShiftExpSumKernel &&) = default; /** Set the input and output tensors. * - * @param[in] input Source tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32 + * @param[in] input Source tensor. Data types supported: QASYMM8/F16/F32 * @param[in] max Max values tensor. Data types supported: same as @p input * @param[out] output Destination tensor. Data types supported: S32 for QASYMM8 @p input, or same as @p input * @param[out] sum Sum of 1D logits tensor. Data types supported: S32 for QASYMM8 @p input, or same as @p input @@ -77,7 +77,7 @@ public: void configure(const ICLTensor *input, const ICLTensor *max, ICLTensor *output, ICLTensor *sum, float beta = 1.0f); /** Static function to check if given info will lead to a valid configuration of @ref CLLogits1DShiftExpSumKernel * - * @param[in] input Source tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32 + * @param[in] input Source tensor. Data types supported: QASYMM8/F16/F32 * @param[in] max Max values tensor. Data types supported: same as @p input * @param[in] output Destination tensor. Data types supported: S32 for QASYMM8 @p input, or same as @p input * @param[in] sum Sum of 1D logits tensor. Data types supported: S32 for QASYMM8 @p input, or same as @p input @@ -116,7 +116,7 @@ public: CLLogits1DMaxShiftExpSumKernel &operator=(CLLogits1DMaxShiftExpSumKernel &&) = default; /** Set the input and output tensors. * - * @param[in] input Source tensor. Data types supported: QS8/QS16/F16/F32 + * @param[in] input Source tensor. Data types supported: F16/F32 * @param[in,out] max Max values tensor. Data types supported: same as @p input * @param[out] output Destination tensor. Data types supported: same as @p input * @param[out] sum Sum of 1D logits tensor. Data types supported: same as @p input @@ -125,7 +125,7 @@ public: void configure(const ICLTensor *input, ICLTensor *max, ICLTensor *output, ICLTensor *sum, float beta = 1.0f); /** Static function to check if given info will lead to a valid configuration of @ref CLLogits1DMaxShiftExpSumKernel * - * @param[in] input Source tensor. Data types supported: QS8/QS16/F16/F32 + * @param[in] input Source tensor. Data types supported: F16/F32 * @param[in] max Max values tensor. Data types supported: same as @p input * @param[in] output Destination tensor. Data types supported: same as @p input * @param[in] sum Sum of 1D logits tensor. Data types supported: same as @p input @@ -175,7 +175,7 @@ public: CLLogits1DNormKernel &operator=(CLLogits1DNormKernel &&) = default; /** Set the input and output tensors. * - * @param[in] input Source tensor. Data types supported: QS8/QS16/S32/F16/F32 + * @param[in] input Source tensor. Data types supported: S32/F16/F32 * @param[in] sum Sum tensor. Dimensions should be dim(input)-1. Data types supported: same as @p input * @param[out] output Destination tensor. Data types supported: QASYMM8 for S32 @p input, or same as @p input * @param[in] beta (Optional) A scaling factor for the exponent. (Default = 1.0) @@ -183,7 +183,7 @@ public: void configure(const ICLTensor *input, const ICLTensor *sum, ICLTensor *output, float beta = 1.0f); /** Static function to check if given info will lead to a valid configuration of @ref CLLogits1DNormKernel * - * @param[in] input Source tensor. Data types supported: QS8/QS16/S32/F16/F32 + * @param[in] input Source tensor. Data types supported: S32/F16/F32 * @param[in] sum Sum tensor. Dimensions should be dim(input)-1. Data types supported: same as @p input * @param[in] output Destination tensor. Data types supported: QASYMM8 for S32 @p input, or same as @p input * diff --git a/arm_compute/core/CL/kernels/CLTransposeKernel.h b/arm_compute/core/CL/kernels/CLTransposeKernel.h index 2e1b481d3f..09d7a8a430 100644 --- a/arm_compute/core/CL/kernels/CLTransposeKernel.h +++ b/arm_compute/core/CL/kernels/CLTransposeKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -40,13 +40,13 @@ class CLTransposeKernel : public ICLSimple2DKernel public: /** Initialise the kernel's input and output. * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[out] output Output tensor. Data type supported: Same as @p input */ void configure(const ICLTensor *input, ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLTransposeKernel * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] output Output tensor. Data type supported: Same as @p input * * @return a status diff --git a/arm_compute/core/CL/kernels/CLWeightsReshapeKernel.h b/arm_compute/core/CL/kernels/CLWeightsReshapeKernel.h index 7a54284199..664fc3c216 100644 --- a/arm_compute/core/CL/kernels/CLWeightsReshapeKernel.h +++ b/arm_compute/core/CL/kernels/CLWeightsReshapeKernel.h @@ -69,7 +69,7 @@ public: /** Set the input and output of the kernel. * * @param[in] input The input tensor to convert. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM] if shared, - * and 5D tensor with dimensions [kernel_x, kernel_y, IFM, OFM, num_patches] if unshared. Data types supported: QS8/QS16/QASYMM8/F16/F32 + * and 5D tensor with dimensions [kernel_x, kernel_y, IFM, OFM, num_patches] if unshared. Data types supported: QASYMM8/F16/F32 * @param[in] biases The shared biases tensor to append. Bias is 1D tensor with dimensions [OFM] if shared and 2D tensor with * dimensions [OFM, num_patches] if unshared. Data types supported: Same as @p input * @warning Appending biases to weights reshaped matrix is not supported for quantized asymmetric types. @@ -79,7 +79,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLWeightsReshapeKernel * * @param[in] input The input tensor to convert. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM] if shared, - * and 5D tensor with dimensions [kernel_x, kernel_y, IFM, OFM, num_patches] if unshared. Data types supported: QS8/QS16/QASYMM8/F16/F32 + * and 5D tensor with dimensions [kernel_x, kernel_y, IFM, OFM, num_patches] if unshared. Data types supported: QASYMM8/F16/F32 * @param[in] biases The shared biases tensor to append. Bias is 1D tensor with dimensions [OFM] if shared and 2D tensor with * dimensions [OFM, num_patches] if unshared. Data types supported: Same as @p input * @warning Appending biases to weights reshaped matrix is not supported for quantized asymmetric types. diff --git a/arm_compute/core/CL/kernels/CLWidthConcatenateLayerKernel.h b/arm_compute/core/CL/kernels/CLWidthConcatenateLayerKernel.h index 5b8a318320..d206eb0da7 100644 --- a/arm_compute/core/CL/kernels/CLWidthConcatenateLayerKernel.h +++ b/arm_compute/core/CL/kernels/CLWidthConcatenateLayerKernel.h @@ -52,7 +52,7 @@ public: ~CLWidthConcatenateLayerKernel() = default; /** Initialise the kernel's inputs and output * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] width_offset The offset on the X axis. * @param[in,out] output Output tensor. Data types supported: Same as @p input. * @@ -60,7 +60,7 @@ public: void configure(const ICLTensor *input, unsigned int width_offset, ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLDepthConcatenateLayerKernel * - * @param[in] input Input tensor info. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input Input tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] width_offset The offset on the X axis. * @param[in] output Output tensor info. Data types supported: Same as @p input. * |