diff options
Diffstat (limited to 'arm_compute/runtime')
53 files changed, 161 insertions, 174 deletions
diff --git a/arm_compute/runtime/CL/functions/CLActivationLayer.h b/arm_compute/runtime/CL/functions/CLActivationLayer.h index 7100eb4d84..e98fa4bf48 100644 --- a/arm_compute/runtime/CL/functions/CLActivationLayer.h +++ b/arm_compute/runtime/CL/functions/CLActivationLayer.h @@ -44,7 +44,7 @@ public: * @note If the output tensor is a nullptr or is equal to the input, 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 parameters. */ @@ -52,7 +52,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLActivationLayer * * @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/runtime/CL/functions/CLArithmeticAddition.h b/arm_compute/runtime/CL/functions/CLArithmeticAddition.h index 9eea846d97..5aba60ad01 100644 --- a/arm_compute/runtime/CL/functions/CLArithmeticAddition.h +++ b/arm_compute/runtime/CL/functions/CLArithmeticAddition.h @@ -33,7 +33,7 @@ class ICLTensor; /** Basic function to run @ref CLArithmeticAdditionKernel * - * @note The tensor data type for the inputs must be U8/QS8/QS16/S16/F16/F32. + * @note The tensor data type for the inputs must be U8/S16/F16/F32. * @note The function performs an arithmetic addition between two tensors. */ class CLArithmeticAddition : public ICLSimpleFunction @@ -41,19 +41,19 @@ class CLArithmeticAddition : public ICLSimpleFunction public: /** Initialise the kernel's inputs, output and convertion policy. * - * @param[in, out] input1 First tensor input. Data types supported: U8/QS8/QASYMM8/QS16/S16/F16/F32. + * @param[in, out] input1 First tensor input. Data types supported: U8/QASYMM8/S16/F16/F32. * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. - * @param[in, out] 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[in, out] input2 Second tensor input. Data types supported: U8, QASYMM8 (only if @p input1 is QASYMM8), S16/F16/F32. * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. - * @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 both inputs are QASYMM8), QS16 (only if both inputs are QS16), S16/F16/F32. + * @param[out] output Output tensor. 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. */ void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, ConvertPolicy policy); /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticAddition * - * @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/runtime/CL/functions/CLArithmeticSubtraction.h b/arm_compute/runtime/CL/functions/CLArithmeticSubtraction.h index 0d3f5bce6a..b9690806d7 100644 --- a/arm_compute/runtime/CL/functions/CLArithmeticSubtraction.h +++ b/arm_compute/runtime/CL/functions/CLArithmeticSubtraction.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -34,7 +34,7 @@ class ICLTensor; /** Basic function to run @ref CLArithmeticSubtractionKernel * - * @note The tensor data type for the inputs must be U8/QS8/QS16/S16/F16/F32. + * @note The tensor data type for the inputs must be U8/S16/F16/F32. * @note The function performs an arithmetic subtraction between two tensors. */ class CLArithmeticSubtraction : public ICLSimpleFunction @@ -42,17 +42,17 @@ class CLArithmeticSubtraction : public ICLSimpleFunction 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 CLArithmeticSubtraction * - * @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/runtime/CL/functions/CLBatchNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h index 9386a86ae5..215046ae7e 100644 --- a/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h +++ b/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h @@ -50,7 +50,7 @@ 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 @@ -65,7 +65,7 @@ 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/runtime/CL/functions/CLChannelShuffleLayer.h b/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h index b632e3b2e3..0bcbfd6276 100644 --- a/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h +++ b/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h @@ -41,14 +41,14 @@ class CLChannelShuffleLayer : public ICLSimpleFunction public: /** Initialize the function * - * @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/runtime/CL/functions/CLConvertFullyConnectedWeights.h b/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h index d8eef34c62..ae0c9d6459 100644 --- a/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h +++ b/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h @@ -37,7 +37,7 @@ class CLConvertFullyConnectedWeights : public ICLSimpleFunction public: /** Initialize the function. * - * @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. @@ -45,7 +45,7 @@ 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 CLConvertFullyConnectedWeights * - * @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/runtime/CL/functions/CLConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h index 82bb71ca6c..c9a74f2a4c 100644 --- a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h @@ -49,7 +49,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -68,7 +68,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported:Same as @p input. * @param[in] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs. @@ -88,7 +88,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] output Destination 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. diff --git a/arm_compute/runtime/CL/functions/CLCopy.h b/arm_compute/runtime/CL/functions/CLCopy.h index d76f0702af..170dc9a613 100644 --- a/arm_compute/runtime/CL/functions/CLCopy.h +++ b/arm_compute/runtime/CL/functions/CLCopy.h @@ -38,7 +38,7 @@ class CLCopy : public ICLSimpleFunction public: /** Initialise the function's source and destination. * - * @param[in] input Source tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32. + * @param[in] input Source tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32. * @param[out] output Output tensor. Data types supported: Same as @p input. * */ diff --git a/arm_compute/runtime/CL/functions/CLDepthConcatenateLayer.h b/arm_compute/runtime/CL/functions/CLDepthConcatenateLayer.h index 00b3b66c97..d505814e73 100644 --- a/arm_compute/runtime/CL/functions/CLDepthConcatenateLayer.h +++ b/arm_compute/runtime/CL/functions/CLDepthConcatenateLayer.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: CLDepthConcatenateLayer(); /** Initialise the kernel's inputs vector and output. * - * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: QS8/QS16/F16/F32. + * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: F16/F32. * @param[out] output Output tensor. Data types supported: Same as @p input. */ void configure(std::vector<ICLTensor *> inputs_vector, ICLTensor *output); diff --git a/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h b/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h index c84dc15508..00fa0a686d 100644 --- a/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h +++ b/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -43,20 +43,16 @@ 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: QS8/U8/U16/S16/Q16/U32/S32/F32. - * @param[out] output The output tensor. Data types supported: QS8/U8/U16/S16/QS16/U32/S32/F32. + * @param[in] input The input tensor to convert. Data types supported: U8/U16/S16/Q16/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. - * It is not used on fixed point conversion. */ void configure(const ICLTensor *input, ICLTensor *output, ConvertPolicy policy, uint32_t shift); }; diff --git a/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h index 4363224a08..31683c51f9 100644 --- a/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h @@ -47,7 +47,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -61,7 +61,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported:Same as @p input. * @param[in] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs. diff --git a/arm_compute/runtime/CL/functions/CLFillBorder.h b/arm_compute/runtime/CL/functions/CLFillBorder.h index 80a8cf20e3..f4515b6700 100644 --- a/arm_compute/runtime/CL/functions/CLFillBorder.h +++ b/arm_compute/runtime/CL/functions/CLFillBorder.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -38,7 +38,7 @@ class CLFillBorder : public ICLSimpleFunction public: /** Initialize the function * - * @param[in,out] tensor Source tensor. Data types supported: QS8/U8/S16/QS16 + * @param[in,out] tensor Source tensor. Data types supported: U8/S16 * @param[in] border_width The border width * @param[in] border_mode Strategy to use for borders. * @param[in] constant_border_value (Optional) Constant value to use for borders if border_mode is set to CONSTANT. diff --git a/arm_compute/runtime/CL/functions/CLFlattenLayer.h b/arm_compute/runtime/CL/functions/CLFlattenLayer.h index e19b0e465f..88df4a7f96 100644 --- a/arm_compute/runtime/CL/functions/CLFlattenLayer.h +++ b/arm_compute/runtime/CL/functions/CLFlattenLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -41,7 +41,7 @@ class CLFlattenLayer : public ICLSimpleFunction public: /** Initialise the kernel's input and output. * - * @param[in] input First input tensor to flatten with at least 3 dimensions. The dimensions over the third will be interpreted as batches. Data types supported: QS8/QS16/F16/F32 + * @param[in] input First input tensor to flatten with at least 3 dimensions. The dimensions over the third will be interpreted as batches. Data types supported: F16/F32 * @param[out] output Output tensor with shape [w*h*d, input_batches] where: * w = width input tensor, h = height input tensor and d = depth input tensor. Data type supported: same as @p input */ diff --git a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h index 127d8acf10..3357868968 100644 --- a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h +++ b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h @@ -48,13 +48,13 @@ class CLFullyConnectedLayerReshapeWeights : public ICLSimpleFunction public: /** Set the input and output tensors. * - * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: QASYMM8/F16/F32. * @param[out] output Destination tensor which stores the transposed input 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 CLFullyConnectedLayerReshapeWeights * - * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: QASYMM8/F16/F32. * @param[in] output Destination tensor which stores the transposed input tensor. Data type supported: Same as @p input. * * @return a status @@ -86,7 +86,7 @@ public: CLFullyConnectedLayer &operator=(CLFullyConnectedLayer &&) = default; /** Set the input and output tensors. * - * @param[in] input Source tensor. Data type supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Source tensor. Data type supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. The weights must be 2 dimensional. Data type supported: Same as @p input * @param[in] biases Bias tensor. It can be nullptr. Data type supported:Same as @p input. * @param[out] output Destination tensor. Data type supported: Same as @p input. @@ -99,7 +99,7 @@ public: bool retain_internal_weights = false); /** Static function to check if given info will lead to a valid configuration of @ref CLFullyConnectedLayer * - * @param[in] input Source tensor. Data type supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Source tensor. Data type supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. The weights must be 2 dimensional. Data type supported: Same as @p input * @param[in] biases Bias tensor. It can be nullptr. Data type supported:Same as @p input. * @param[in] output Destination tensor. Data type supported: Same as @p input. diff --git a/arm_compute/runtime/CL/functions/CLGEMM.h b/arm_compute/runtime/CL/functions/CLGEMM.h index 41c7467a3f..c4513f29d9 100644 --- a/arm_compute/runtime/CL/functions/CLGEMM.h +++ b/arm_compute/runtime/CL/functions/CLGEMM.h @@ -69,7 +69,7 @@ public: * * @note Whilst the first input tensor can be a vector, the second input tensor must be at least a matrix * - * @param[in] a First input tensor (Matrix or Vector A). Data types supported: QS8/QS16/F16/F32 + * @param[in] a First input tensor (Matrix or Vector A). Data types supported: F16/F32 * @param[in] b Second input tensor (Matrix B). Data type supported: same as @p a. * @param[in] c Third input tensor (Matrix C). It can be a nullptr if just the multiplication between @p a and @p b is needed. Data type supported: same as @p a. * @param[out] output Output tensor. Data type supported: same as @p a @@ -82,7 +82,7 @@ public: void configure(const ICLTensor *a, const ICLTensor *b, const ICLTensor *c, ICLTensor *output, float alpha, float beta, const GEMMInfo &gemm_info = GEMMInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMM. * - * @param[in] a First input tensor info (Matrix or Vector A). Data types supported: QS8/QS16/F16/F32 + * @param[in] a First input tensor info (Matrix or Vector A). Data types supported: F16/F32 * @param[in] b Second input tensor info (Matrix B). Data type supported: same as @p a. * @param[in] c Third input tensor info (Matrix C). It can be a nullptr if just the multiplication between @p a and @p b is needed. Data type supported: same as @p a. * @param[out] output Output tensor info. Data type supported: same as @p a diff --git a/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h index 2c1f7a9d5e..09daa5f568 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h @@ -59,7 +59,7 @@ public: /** Set the input and output tensors. * * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. - * Data type supported: QS8/QASYMM8/QS16/F16/F32. + * Data type supported: QASYMM8/F16/F32. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported: Same as @p weights. * @param[out] output Destination tensor. Data types supported: Same as @p weights. */ @@ -67,7 +67,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLConvolutionLayerReshapeWeights * * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. - * Data type supported: QS8/QASYMM8/QS16/F16/F32. + * Data type supported: QASYMM8/F16/F32. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported: Same as @p weights. * @param[in] output Destination tensor. Data types supported: Same as @p weights. * @@ -116,7 +116,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -134,7 +134,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -158,7 +158,7 @@ public: private: /** Configures the appropriate matrix multiply routine * - * @param[in] input Input tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Input tensor. Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Data type supported: Same as @p input. * @param[in, out] output Output tensor. Data types supported: Same as @p input, * except for input of QASYMM8 type where output should be of S32 type. @@ -167,7 +167,7 @@ private: void configure_mm(const ICLTensor *input, const ICLTensor *weights, ICLTensor *output, int gemm_3d_depth = 1); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMConvolutionLayer matrix multiply routines * - * @param[in] input Input tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Input tensor. Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Data type supported: Same as @p input. * @param[in] output Output tensor. Data types supported: Same as @p input, * except for input of QASYMM8 type where output should be of S32 type. diff --git a/arm_compute/runtime/CL/functions/CLGEMMInterleave4x4.h b/arm_compute/runtime/CL/functions/CLGEMMInterleave4x4.h index ae05b0fd9c..48bb6ccb22 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMInterleave4x4.h +++ b/arm_compute/runtime/CL/functions/CLGEMMInterleave4x4.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -40,7 +40,7 @@ class CLGEMMInterleave4x4 : public ICLSimpleFunction public: /** Initialise the kernel's inputs, output * - * @param[in] input First input tensor. Data types supported: U8/S8/QS8/QASYMM8/QS16/U16/S16/F16/U32/S32/F32 + * @param[in] input First 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); diff --git a/arm_compute/runtime/CL/functions/CLGEMMTranspose1xW.h b/arm_compute/runtime/CL/functions/CLGEMMTranspose1xW.h index ae56548c27..13e643a77d 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMTranspose1xW.h +++ b/arm_compute/runtime/CL/functions/CLGEMMTranspose1xW.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -38,7 +38,7 @@ class CLGEMMTranspose1xW : public ICLSimpleFunction public: /** Initialise the kernel's inputs, output * - * @param[in] input First input tensor. Data type supported: U8/S8/QS8/QASYMM8/QS16/U16/S16/F16/U32/S32/F32 + * @param[in] input First input tensor. Data type 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); diff --git a/arm_compute/runtime/CL/functions/CLNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLNormalizationLayer.h index 93925778d7..89e20d20f6 100644 --- a/arm_compute/runtime/CL/functions/CLNormalizationLayer.h +++ b/arm_compute/runtime/CL/functions/CLNormalizationLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -51,7 +51,7 @@ public: /** Set the input and output tensors. * * @param[in, out] 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 (Written to by the border handler) + * and an optional 4th dimension for batch of inputs. Data types supported: F16/F32 (Written to by the border handler) * @param[out] output Destination tensor. Dimensions, data type and number of channels must match the input ones. * @param[in] norm_info Normalization layer information like the normalization type, normalization size and other parameters. */ @@ -59,7 +59,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref CLNormalizationLayer * * @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. Dimensions, data type and number of channels must match the input ones. * @param[in] norm_info Normalization layer information like the normalization type, normalization size and other parameters. * diff --git a/arm_compute/runtime/CL/functions/CLPermute.h b/arm_compute/runtime/CL/functions/CLPermute.h index a41567570a..638207fc48 100644 --- a/arm_compute/runtime/CL/functions/CLPermute.h +++ b/arm_compute/runtime/CL/functions/CLPermute.h @@ -39,14 +39,14 @@ class CLPermute : public ICLSimpleFunction public: /** Set the input and output tensors. * - * @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 CLPermute. * - * @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/runtime/CL/functions/CLPixelWiseMultiplication.h b/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h index 75b67cd17c..a59fb4aba8 100644 --- a/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h +++ b/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h @@ -37,13 +37,13 @@ class CLPixelWiseMultiplication : public ICLSimpleFunction public: /** Initialise the kernel's inputs, output and convertion policy. * - * @param[in, out] input1 An input tensor. Data types supported: U8/QS8/QS16/S16/F16/F32. + * @param[in, out] input1 An input tensor. Data types supported: U8/S16/F16/F32. * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. * @param[in, out] input2 An input tensor. Data types supported: same as @p input1. * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. - * @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. */ @@ -51,11 +51,11 @@ public: ConvertPolicy overflow_policy, RoundingPolicy rounding_policy); /** Static function to check if given info will lead to a valid configuration of @ref CLPixelWiseMultiplication * - * @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/runtime/CL/functions/CLPoolingLayer.h b/arm_compute/runtime/CL/functions/CLPoolingLayer.h index a8bdabad98..f7571c1d2d 100644 --- a/arm_compute/runtime/CL/functions/CLPoolingLayer.h +++ b/arm_compute/runtime/CL/functions/CLPoolingLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -43,14 +43,14 @@ class CLPoolingLayer : public ICLSimpleFunction public: /** Set the input and output tensors. * - * @param[in,out] input Source tensor. (Written to only when padding != 0) Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in,out] input Source tensor. (Written to only when padding != 0) 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(ICLTensor *input, ICLTensor *output, const PoolingLayerInfo &pool_info); /** Static function to check if given info will lead to a valid configuration of @ref CLPoolingLayer * - * @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/runtime/CL/functions/CLReshapeLayer.h b/arm_compute/runtime/CL/functions/CLReshapeLayer.h index 411e751290..cf5f7e5e15 100644 --- a/arm_compute/runtime/CL/functions/CLReshapeLayer.h +++ b/arm_compute/runtime/CL/functions/CLReshapeLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -36,7 +36,7 @@ class CLReshapeLayer : public ICLSimpleFunction public: /** Initialise the kernel's inputs and outputs * - * @param[in] input First tensor input. Data type supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/U32/S32/F16/F32 + * @param[in] input First tensor input. Data type supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32 * @param[out] output Output tensor. Data type supported: Same as @p input */ void configure(const ICLTensor *input, ICLTensor *output); diff --git a/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h b/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h index 8ab6c160d1..34349ed52b 100644 --- a/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h +++ b/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h @@ -53,14 +53,14 @@ public: CLSoftmaxLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr); /** 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 * @param[in] beta (Optional) A scaling factor for the exponent. Defaults to 1.f */ void configure(const ICLTensor *input, ICLTensor *output, float beta = 1.0f); /** Static function to check if given info will lead to a valid configuration of @ref CLSoftmaxLayer * - * @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 diff --git a/arm_compute/runtime/CL/functions/CLTranspose.h b/arm_compute/runtime/CL/functions/CLTranspose.h index 89a2022e76..1e9bb95114 100644 --- a/arm_compute/runtime/CL/functions/CLTranspose.h +++ b/arm_compute/runtime/CL/functions/CLTranspose.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 CLTranspose : public ICLSimpleFunction public: /** Initialise the kernel's inputs and output * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/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 CLTranspose * - * @param[in] input The input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/F16/U32/S32/F32 + * @param[in] input The input tensor. 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 * * @return a status diff --git a/arm_compute/runtime/CL/functions/CLWidthConcatenateLayer.h b/arm_compute/runtime/CL/functions/CLWidthConcatenateLayer.h index bcda05274b..289191e030 100644 --- a/arm_compute/runtime/CL/functions/CLWidthConcatenateLayer.h +++ b/arm_compute/runtime/CL/functions/CLWidthConcatenateLayer.h @@ -50,13 +50,13 @@ public: CLWidthConcatenateLayer(); /** Initialise the kernel's inputs vector and output. * - * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @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. */ void configure(std::vector<ICLTensor *> inputs_vector, ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLDepthConcatenateLayerKernel * - * @param[in] inputs_vector The vectors containing all the tensors info to concatenate. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @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. * * @return a status diff --git a/arm_compute/runtime/CPP/functions/CPPPermute.h b/arm_compute/runtime/CPP/functions/CPPPermute.h index 0094576da6..40d6830425 100644 --- a/arm_compute/runtime/CPP/functions/CPPPermute.h +++ b/arm_compute/runtime/CPP/functions/CPPPermute.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -38,14 +38,14 @@ class CPPPermute : public ICPPSimpleFunction public: /** Configure the permute CPP 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[out] output The output tensor. Data types supported: Same as @p input * @param[in] perm Permutation vector */ void configure(const ITensor *input, ITensor *output, const PermutationVector &perm); /** Static function to check if given info will lead to a valid configuration of @ref CPPPermute * - * @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 * diff --git a/arm_compute/runtime/GLES_COMPUTE/functions/GCConvolutionLayer.h b/arm_compute/runtime/GLES_COMPUTE/functions/GCConvolutionLayer.h index 45a883948c..421150e18e 100644 --- a/arm_compute/runtime/GLES_COMPUTE/functions/GCConvolutionLayer.h +++ b/arm_compute/runtime/GLES_COMPUTE/functions/GCConvolutionLayer.h @@ -108,7 +108,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -139,7 +139,7 @@ private: void configure_mm(const IGCTensor *input, const IGCTensor *weights, IGCTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref GCGEMMConvolutionLayer matrix multiply routines * - * @param[in] input Input tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Input tensor. Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Data type supported: Same as @p input. * @param[in] output Output tensor. Data types supported: Same as @p input, * except for input of QASYMM8 type where output should be of S32 type. diff --git a/arm_compute/runtime/NEON/functions/NEActivationLayer.h b/arm_compute/runtime/NEON/functions/NEActivationLayer.h index 59f5802d2a..a65146d461 100644 --- a/arm_compute/runtime/NEON/functions/NEActivationLayer.h +++ b/arm_compute/runtime/NEON/functions/NEActivationLayer.h @@ -44,7 +44,7 @@ public: * @note If the output tensor is a nullptr or is equal to the input, 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] activation_info Activation layer parameters. */ @@ -52,7 +52,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref NEActivationLayer * * @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/runtime/NEON/functions/NEArithmeticAddition.h b/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h index c72d0b6d61..c29646397c 100644 --- a/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h +++ b/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h @@ -37,17 +37,17 @@ class NEArithmeticAddition : public INESimpleFunction public: /** Initialise the kernel's inputs, output and conversion 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/QS16/S16/F16/F32 - * @param[out] output Output tensor. Data types supported: U8/QS8/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/S16/F16/F32 * @param[in] policy Policy to use to handle overflow. */ void configure(ITensor *input1, ITensor *input2, ITensor *output, ConvertPolicy policy); /** Static function to check if given info will lead to a valid configuration of @ref NEArithmeticAddition * - * @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/QS16/S16/F16/F32 - * @param[in] output Output tensor. Data types supported: U8/QS8/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[in] output Output tensor. Data types supported: U8/S16/F16/F32 * @param[in] policy Policy to use to handle overflow. * * @return a status diff --git a/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h b/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h index 751ed1adf1..9b460c1031 100644 --- a/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h +++ b/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -37,17 +37,17 @@ class NEArithmeticSubtraction : public INESimpleFunction public: /** Initialise the kernel's inputs, output and conversion 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/QS16/S16/F16/F32 - * @param[out] output Output tensor. Data types supported: U8/QS8/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/S16/F16/F32 * @param[in] policy Policy to use to handle overflow. */ void configure(const ITensor *input1, const ITensor *input2, ITensor *output, ConvertPolicy policy); /** Static function to check if given info will lead to a valid configuration of @ref NEArithmeticSubtraction * - * @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/QS16/S16/F16/F32 - * @param[in] output Output tensor. Data types supported: U8/QS8/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[in] output Output tensor. Data types supported: U8/S16/F16/F32 * @param[in] policy Policy to use to handle overflow. * * @return a status diff --git a/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h b/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h index feb2087aa0..77f06129a3 100644 --- a/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h +++ b/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h @@ -50,7 +50,7 @@ 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 @@ -65,7 +65,7 @@ public: * * @param[in] input Source tensor info. 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[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/runtime/NEON/functions/NECol2Im.h b/arm_compute/runtime/NEON/functions/NECol2Im.h index 9b05bd4513..42876a8aec 100644 --- a/arm_compute/runtime/NEON/functions/NECol2Im.h +++ b/arm_compute/runtime/NEON/functions/NECol2Im.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -39,7 +39,7 @@ class NECol2Im : public INESimpleFunction public: /** Configure the col2im NEON kernel * - * @param[in] input The input tensor to convert. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input The input tensor to convert. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/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. @@ -47,7 +47,7 @@ public: void configure(const ITensor *input, ITensor *output, const Size2D &convolved_dims); /** Static function to check if given info will lead to a valid configuration of @ref NECol2Im * - * @param[in] input The input tensor to convert. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32 + * @param[in] input The input tensor to convert. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/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/runtime/NEON/functions/NEConvertFullyConnectedWeights.h b/arm_compute/runtime/NEON/functions/NEConvertFullyConnectedWeights.h index bdb157f30b..3ec0390124 100644 --- a/arm_compute/runtime/NEON/functions/NEConvertFullyConnectedWeights.h +++ b/arm_compute/runtime/NEON/functions/NEConvertFullyConnectedWeights.h @@ -40,7 +40,7 @@ public: NEConvertFullyConnectedWeights(); /** Initialize the function. * - * @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. @@ -48,7 +48,7 @@ public: void configure(const ITensor *input, ITensor *output, const TensorShape &original_input_shape, DataLayout data_layout); /** Static function to check if given info will lead to a valid configuration of @ref NEConvertFullyConnectedWeights * - * @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/runtime/NEON/functions/NEConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h index e143814a4e..c4226cbc5d 100644 --- a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h @@ -52,7 +52,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -72,7 +72,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -94,7 +94,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] output Destination 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. diff --git a/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h b/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h index 5b63b70634..eefb5fa362 100644 --- a/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.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: NEDepthConcatenateLayer(); /** Initialise the kernel's inputs vector and output. * - * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: QS8/QS16/F16/F32. + * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: F16/F32. * @param[out] output Output tensor. Data types supported: Same as @p inputs_vector. */ void configure(std::vector<ITensor *> inputs_vector, ITensor *output); diff --git a/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h b/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h index b235e87b4a..eedadc242d 100644 --- a/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -46,22 +46,14 @@ public: /** Initialize the function's source, destination * * Valid conversions Input -> Output : - * QS8 -> QS8, F32 * U8 -> U16, S16, S32 * U16 -> U8, U32 * S16 -> U8, S32 - * QS16 -> QS16, F32 - * F32 -> QS8, QS16 * - * @warning In case of in-place fixed point position conversion make sure that configure has been called - * before the updated tensor is used in other functions, as the TensorInfo of the tensor will be - * altered. In-place is only supported for QS8 -> QS8, QS16 -> QS16. - * - * @param[in, out] input The input tensor to convert (Written in case of in-place computation). Data types supported: U8/QS8/U16/S16/F32. - * @param[out] output The output tensor. Can be null in case of in-place computation. Data types supported: U8/QS8/U16/S16/U32/S32/F32. + * @param[in, out] input The input tensor to convert (Written in case of in-place computation). Data types supported: U8/U16/S16/F32. + * @param[out] output The output tensor. Can be null in case of in-place computation. Data types supported: U8/U16/S16/U32/S32/F32. * @param[in] policy Conversion policy. * @param[in] shift (Optional) Value for down/up conversions. Must be 0 <= shift < 8. - * In case of fixed point position conversion, it specifies the new fixed point position, if operation is in-place. */ void configure(ITensor *input, ITensor *output, ConvertPolicy policy, uint32_t shift = 0); }; diff --git a/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h index ae384ffa56..a4a55d10f8 100644 --- a/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h @@ -54,11 +54,11 @@ public: /** Set the input, weights, biases and output tensors. * * @note: DirectConvolution only works in the following configurations: - * 1x1 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = QS8/QS16/F16/F32 - * 3x3 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = QS8/F16/F32 + * 1x1 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F16/F32 + * 3x3 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F16/F32 * 5x5 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F32 * - * @param[in, out] input Input tensor. Data types supported: QS8/QS16/F16/F32. + * @param[in, out] input Input tensor. Data types supported: F16/F32. * @param[in] weights Set of kernels to convolve the input volume. * Supported sizes: 1x1, 3x3 and 5x5. * The 3rd dimension must be the same as the input's volume 3rd dimension. @@ -73,11 +73,11 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref NEDirectConvolutionLayer * * @note: DirectConvolution only works in the following configurations: - * 1x1 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = QS8/QS16/F16/F32 - * 3x3 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = QS8/F16/F32 + * 1x1 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F16/F32 + * 3x3 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F16/F32 * 5x5 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F32 * - * @param[in] input Input tensor. Data types supported: QS8/QS16/F16/F32. + * @param[in] input Input tensor. Data types supported: F16/F32. * @param[in] weights Set of kernels to convolve the input volume. * Supported sizes: 1x1, 3x3 and 5x5. * The 3rd dimension must be the same as the input's volume 3rd dimension. @@ -104,7 +104,6 @@ private: NEActivationLayer _activationlayer_function; Tensor _accumulator; bool _has_bias; - bool _is_fixed_point; bool _is_activationlayer_enabled; unsigned int _dim_split; }; diff --git a/arm_compute/runtime/NEON/functions/NEFillBorder.h b/arm_compute/runtime/NEON/functions/NEFillBorder.h index b6b7e77471..27a9eea9af 100644 --- a/arm_compute/runtime/NEON/functions/NEFillBorder.h +++ b/arm_compute/runtime/NEON/functions/NEFillBorder.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016, 2017 ARM Limited. + * Copyright (c) 2016-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -41,7 +41,7 @@ public: * * @note This function fills the borders within the XY-planes. * - * @param[in, out] input Source tensor. Data type supported: U8/QS8/S16/S32/F32 + * @param[in, out] input Source tensor. Data type supported: U8/S16/S32/F32 * @param[in] border_width Width of the tensor border in pixels. * @param[in] border_mode Strategy to use for borders. * @param[in] constant_border_value (Optional) Constant value to use for borders if border_mode is set to CONSTANT. diff --git a/arm_compute/runtime/NEON/functions/NEFlattenLayer.h b/arm_compute/runtime/NEON/functions/NEFlattenLayer.h index e9c8e27d57..2c259fa178 100644 --- a/arm_compute/runtime/NEON/functions/NEFlattenLayer.h +++ b/arm_compute/runtime/NEON/functions/NEFlattenLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -41,7 +41,7 @@ class NEFlattenLayer : public INESimpleFunction public: /** Initialise the kernel's input and output. * - * @param[in] input First input tensor to flatten with at least 3 dimensions. The dimensions over the third will be interpreted as batches. Data types supported: QS8/QS16/F16/F32 + * @param[in] input First input tensor to flatten with at least 3 dimensions. The dimensions over the third will be interpreted as batches. Data types supported: F16/F32 * @param[out] output Output tensor with shape [w*h*d, input_batches] where: * w = width input tensor, h = height input tensor and d = depth input tensor. Data type supported: same as @p input */ diff --git a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h index 42c9e2d3e9..d4166b3830 100644 --- a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h +++ b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h @@ -51,7 +51,7 @@ public: NEFullyConnectedLayerReshapeWeights(std::shared_ptr<IMemoryManager> memory_manager = nullptr); /** Set the input and output tensors. * - * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: QS8/QS16/F32. + * @param[in] input Weights tensor. The weights must be 2 dimensional. Data types supported: F32. * @param[out] output Destination tensor. Data type supported: Same as @p input. * @param[in] transpose_weights True if the weights must be transposed. Data types supported: Same as @p weights. * @param[in] is_batched_fc_layer True if it is a batched fully connected layer @@ -59,7 +59,7 @@ public: void configure(const ITensor *input, ITensor *output, bool transpose_weights, bool is_batched_fc_layer); /** Static function to check if given info will lead to a valid configuration of @ref CLFullyConnectedLayerReshapeWeights * - * @param[in] input Weights tensor info. The weights must be 2 dimensional. Data types supported: QS8/QS16/F32. + * @param[in] input Weights tensor info. The weights must be 2 dimensional. Data types supported: F32. * @param[in] output Destination tensor info. Data type supported: Same as @p input. * @param[in] transpose_weights True if the weights must be transposed. Data types supported: Same as @p weights. * @param[in] is_batched_fc_layer True if it is a batched fully connected layer @@ -104,7 +104,7 @@ public: NEFullyConnectedLayer &operator=(NEFullyConnectedLayer &&) = default; /** Set the input and output tensors. * - * @param[in] input Source tensor. Data type supported: QS8/QS16/F16/F32. + * @param[in] input Source tensor. Data type supported: F16/F32. * @param[in] weights Weights tensor. The weights must be 2 dimensional. Data type supported: Same as @p input. * @param[in] biases Bias tensor. Can be nullptr. Data type supported:Same as @p input. * @param[out] output Destination tensor. Data type supported: Same as @p input. @@ -114,7 +114,7 @@ public: void configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, bool transpose_weights = true, bool are_weights_reshaped = false); /** Static function to check if given info will lead to a valid configuration of @ref CLFullyConnectedLayer * - * @param[in] input Source tensor info. Data type supported: QS8/QS16/F16/F32. + * @param[in] input Source tensor info. Data type supported: F16/F32. * @param[in] weights Weights tensor info. The weights must be 2 dimensional. Data type supported: Same as @p input * @param[in] biases Bias tensor info. It can be nullptr. Data type supported:Same as @p input. * @param[in] output Destination tensor info. Data type supported: Same as @p input. diff --git a/arm_compute/runtime/NEON/functions/NEGEMM.h b/arm_compute/runtime/NEON/functions/NEGEMM.h index 5d108b2c14..cf059e5c4d 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMM.h +++ b/arm_compute/runtime/NEON/functions/NEGEMM.h @@ -66,7 +66,7 @@ public: * @note GEMM: General Matrix Multiply - [alpha * A * B + beta * C]. * @note GEMM: The tensors a, b, c, d must have the same data type. You should not mix data types when calling this function. * - * @param[in] a First input tensor (Matrix A or Vector A). Data type supported: QS8/QS16/F16/F32 + * @param[in] a First input tensor (Matrix A or Vector A). Data type supported: F16/F32 * @param[in] b Second input tensor (Matrix B). Data type supported: same as @p a * @param[in] c Third input tensor (Matrix C). It can be a nullptr if just the multiplication between @p a and @p b is needed. Data type supported: same as @p a * @param[out] d Output tensor. Data type supported: same as @p a diff --git a/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h index 7075becf75..68e1145e35 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h @@ -60,7 +60,7 @@ public: NEConvolutionLayerReshapeWeights(std::shared_ptr<IMemoryManager> memory_manager = nullptr); /** Set the input and output tensors. * - * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: QS8/QASYMM8/QS16/F32. + * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: QASYMM8/F32. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported: Same as @p weights. * @param[out] output Destination tensor. Data types supported: Same as @p weights. * @param[in] transpose1xW True if the weights are to undergo a 1xW transposition after reshaping (in case of GEMM operation), false otherwise. @@ -69,7 +69,7 @@ public: void configure(const ITensor *weights, const ITensor *biases, ITensor *output, bool transpose1xW); /** Static function to check if given info will lead to a valid configuration of @ref NEConvolutionLayerReshapeWeights * - * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: QASYMM8/F16/F32. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported: Same as @p weights. * @param[in] output Destination tensor. Data types supported: Same as @p weights. * @param[in] transpose1xW True if the weights are to undergo a 1xW transposition after reshaping (in case of GEMM operation), false otherwise. @@ -116,7 +116,7 @@ public: * * @param[in] input Source tensor. 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/F32. + * Data types supported: QASYMM8/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported: Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -134,7 +134,7 @@ public: * * @param[in] input Source tensor. 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. + * Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM]. Data type supported:Same as @p input. * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type. @@ -158,7 +158,7 @@ public: private: /** Configures the appropriate matrix multiply routine * - * @param[in] input Input tensor. Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Input tensor. Data types supported: QASYMM8/F16/F32. * @param[in] weights Weights tensor. Data type supported: Same as @p input. * @param[out] output Output tensor. Data types supported: Same as @p input, * except for input of QASYMM8 type where output should be of S32 type. diff --git a/arm_compute/runtime/NEON/functions/NEGEMMInterleave4x4.h b/arm_compute/runtime/NEON/functions/NEGEMMInterleave4x4.h index b911fd064f..4a6bec03e6 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMInterleave4x4.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMInterleave4x4.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -40,7 +40,7 @@ class NEGEMMInterleave4x4 : public INESimpleFunction public: /** Initialise the kernel's inputs, output * - * @param[in] input First input tensor. Data types supported: U8/S8/QS8/U16/S16/F16/U32/S32/F32 + * @param[in] input First input tensor. Data types supported: U8/S8/U16/S16/F16/U32/S32/F32 * @param[out] output Output tensor. Data type supported: same as @p input */ void configure(const ITensor *input, ITensor *output); diff --git a/arm_compute/runtime/NEON/functions/NEGEMMTranspose1xW.h b/arm_compute/runtime/NEON/functions/NEGEMMTranspose1xW.h index 8b9ad136b4..3f8e731d01 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMTranspose1xW.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMTranspose1xW.h @@ -38,13 +38,13 @@ class NEGEMMTranspose1xW : public INESimpleFunction public: /** Initialise the kernel's inputs, output * - * @param[in] input First input tensor. Data type supported: U8/S8/QS8/U16/S16/F16/U32/S32/F32/ + * @param[in] input First input tensor. Data type supported: U8/S8/U16/S16/F16/U32/S32/F32/ * @param[out] output Output tensor. Data type supported: same as @p input */ void configure(const ITensor *input, ITensor *output); /** Static function to check if given info will lead to a valid configuration of @ref NEGEMMTranspose1xW * - * @param[in] input First input tensor. Data type supported: U8/S8/QS8/U16/S16/F16/U32/S32/F32/ + * @param[in] input First input tensor. Data type supported: U8/S8/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/runtime/NEON/functions/NEIm2Col.h b/arm_compute/runtime/NEON/functions/NEIm2Col.h index caa8a011f6..d888b7e8f5 100644 --- a/arm_compute/runtime/NEON/functions/NEIm2Col.h +++ b/arm_compute/runtime/NEON/functions/NEIm2Col.h @@ -43,7 +43,7 @@ public: /** Configure the im2col NEON 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/QS16/QASYMM8/F16/F32 + * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F16/F32 * Note: QASYMM8 works only for has_bias = false * @param[out] output The output tensor. Data types supported: Same as @p input * @param[in] kernel_dims The kernel dimensions (width and height). @@ -56,7 +56,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref NEIm2Col * * @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/QS16/QASYMM8/F16/F32 + * while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F16/F32 * Note: QASYMM8 works only for has_bias = false * @param[in] output The output tensor. Data types supported: Same as @p input * @param[in] kernel_dims The kernel dimensions (width and height). diff --git a/arm_compute/runtime/NEON/functions/NENormalizationLayer.h b/arm_compute/runtime/NEON/functions/NENormalizationLayer.h index 4b5ad28706..4f1f32fba5 100644 --- a/arm_compute/runtime/NEON/functions/NENormalizationLayer.h +++ b/arm_compute/runtime/NEON/functions/NENormalizationLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -55,7 +55,7 @@ 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 type supported: QS8/QS16/F16/F32 + * and an optional 4th dimension for batch of inputs. Data type supported: F16/F32 * @param[out] output Destination with the same dimensions, data type and number of channels of @p input * @param[in] norm_info Normalization layer information like the normalization type, normalization size and other parameters. */ @@ -63,7 +63,7 @@ public: /** Static function to check if given info will lead to a valid configuration of @ref NENormalizationLayer * * @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 type supported: QS8/QS16/F16/F32 + * and an optional 4th dimension for batch of inputs. Data type supported: F16/F32 * @param[in] output Destination with the same dimensions, data type and number of channels of @p input * @param[in] norm_info Normalization layer information like the normalization type, normalization size and other parameters. * diff --git a/arm_compute/runtime/NEON/functions/NEPermute.h b/arm_compute/runtime/NEON/functions/NEPermute.h index 58626cd2f2..580d24e415 100644 --- a/arm_compute/runtime/NEON/functions/NEPermute.h +++ b/arm_compute/runtime/NEON/functions/NEPermute.h @@ -40,7 +40,7 @@ public: * * @note Supported permutation vectors : [2, 0, 1], [1, 2, 0] * - * @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[out] output The output tensor. Data types supported: Same as @p input * @param[in] perm Permutation vector */ @@ -49,7 +49,7 @@ public: * * @note Supported permutation vectors : [2, 0, 1], [1, 2, 0] * - * @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 * diff --git a/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h b/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h index ba96ae6cfa..371bb2e13e 100644 --- a/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h +++ b/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h @@ -37,22 +37,22 @@ class NEPixelWiseMultiplication : public INESimpleFunction public: /** Initialise the kernel's inputs, output and convertion policy. * - * @param[in, out] input1 An input tensor. Data types supported: U8/QS8/S16/F16/F32. + * @param[in, out] input1 An input tensor. Data types supported: U8/S16/F16/F32. * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. * @param[in, out] input2 An input tensor. Data types supported: same as @p input1. * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. - * @param[out] output Output tensor. Data types supported: U8/QS8/S16/F16/F32. + * @param[out] output Output tensor. Data types supported: U8/S16/F16/F32. * @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. * @param[in] rounding_policy Rounding policy. */ void configure(ITensor *input1, ITensor *input2, ITensor *output, float scale, ConvertPolicy overflow_policy, RoundingPolicy rounding_policy); /** Static function to check if given info will lead to a valid configuration of @ref NEPixelWiseMultiplication * - * @param[in] input1 First tensor info input. Data types supported: U8/QS8/S16/F16/F32. - * @param[in] input2 Second tensor info input. Data types supported: U8/QS8/S16/F16/F32. - * @param[in] output Output tensor info. Data types supported: U8/QS8/S16/F16/F32. + * @param[in] input1 First tensor info input. Data types supported: U8/S16/F16/F32. + * @param[in] input2 Second tensor info input. Data types supported: U8/S16/F16/F32. + * @param[in] output Output tensor info. Data types supported: U8/S16/F16/F32. * @param[in] scale Scale to apply after multiplication. Must be positive. * @param[in] overflow_policy Overflow policy. * @param[in] rounding_policy Rounding policy. diff --git a/arm_compute/runtime/NEON/functions/NEPoolingLayer.h b/arm_compute/runtime/NEON/functions/NEPoolingLayer.h index 4224f75c77..26858d5cde 100644 --- a/arm_compute/runtime/NEON/functions/NEPoolingLayer.h +++ b/arm_compute/runtime/NEON/functions/NEPoolingLayer.h @@ -46,18 +46,18 @@ public: NEPoolingLayer(); /** Set the input and output tensors. * - * @note QS8, QS16 and F16 are supported for pool sizes 2 and 3 only + * @note F16 is supported for pool sizes 2 and 3 only * - * @param[in, out] input Source tensor. (Written to only when padding != 0) Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in, out] input Source tensor. (Written to only when padding != 0) 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(ITensor *input, ITensor *output, const PoolingLayerInfo &pool_info); /** Static function to check if given info will lead to a valid configuration of @ref NEPoolingLayer * - * @note QS8, QS16 and F16 are supported for pool sizes 2 and 3 only + * @note F16 is supported for pool sizes 2 and 3 only * - * @param[in] input Source tensor. (Written to only when padding != 0) Data types supported: QS8/QASYMM8/QS16/F16/F32. + * @param[in] input Source tensor. (Written to only when padding != 0) Data types supported: QASYMM8/F16/F32. * @param[in] output Destination tensor. Data types supported: Same as @p input. * @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo. * diff --git a/arm_compute/runtime/NEON/functions/NEReshapeLayer.h b/arm_compute/runtime/NEON/functions/NEReshapeLayer.h index 0bab534ebc..a77a5f30dc 100644 --- a/arm_compute/runtime/NEON/functions/NEReshapeLayer.h +++ b/arm_compute/runtime/NEON/functions/NEReshapeLayer.h @@ -37,7 +37,7 @@ class NEReshapeLayer : public INESimpleFunction public: /** Initialise the kernel's inputs and outputs * - * @param[in] input First tensor input. Data type supported: U8/S8/QS8/QASYMM8//U16/S16/QS16/U32/S32/F16/F32 + * @param[in] input First tensor input. Data type supported: U8/S8/QASYMM8//U16/S16/U32/S32/F16/F32 * @param[out] output Output tensor. Data type supported: Same as @p input */ void configure(const ITensor *input, ITensor *output); diff --git a/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h b/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h index 3d981b6f75..61f46004d6 100644 --- a/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h +++ b/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h @@ -51,18 +51,18 @@ public: NESoftmaxLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr); /** Set the input and output tensors. * - * @param[in,out] input Source tensor. Data types supported: QASYMM8/QS8/QS16/F16/F32. If the width is not a + * @param[in,out] input Source tensor. Data types supported: QASYMM8/F16/F32. If the width is not a * multiple of the internal processing block size, @ref NEFillBorderKernel replicates the * last value of each row to the nearest multiple. * @param[out] output Destination tensor. Data types supported: same as @p input. - * @param[in] beta (Optional) A scaling factor for the exponent. QS8/QS16 only support a beta value of 1. + * @param[in] beta (Optional) A scaling factor for the exponent. */ void configure(ITensor *input, ITensor *output, float beta = 1.0f); /** Static function to check if given info will lead to a valid configuration of @ref NESoftmaxLayer * - * @param[in] input Source tensor. Data types supported: QASYMM8/QS8/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 - * @param[in] beta (Optional) A scaling factor for the exponent. QS8/QS16 only support a beta value of 1. + * @param[in] beta (Optional) A scaling factor for the exponent. * * @return a status */ diff --git a/arm_compute/runtime/NEON/functions/NETranspose.h b/arm_compute/runtime/NEON/functions/NETranspose.h index 6d1e107084..0234288b4b 100644 --- a/arm_compute/runtime/NEON/functions/NETranspose.h +++ b/arm_compute/runtime/NEON/functions/NETranspose.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -41,13 +41,13 @@ class NETranspose : public INESimpleFunction public: /** Initialise the kernel's inputs and output * - * @param[in] input Input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/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 ITensor *input, ITensor *output); /** Static function to check if given info will lead to a valid configuration of @ref NETranspose * - * @param[in] input The input tensor. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/F16/U32/S32/F32 + * @param[in] input The input tensor. 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 * * @return a status |