Functions
arm_cmsis_nn_status	arm_convolve_1_x_n_s8 (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int32_t bias_data, const cmsis_nn_dims output_dims, int8_t *output_data)
	1xn convolution More...

int32_t	arm_convolve_1_x_n_s8_get_buffer_size (const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims)
	Get the required additional buffer size for 1xn convolution. More...

arm_cmsis_nn_status	arm_convolve_1x1_s8 (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int32_t bias_data, const cmsis_nn_dims output_dims, int8_t *output_data)
	s8 version for 1x1 convolution with support for non-unity stride values More...

arm_cmsis_nn_status	arm_convolve_1x1_s8_fast (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int32_t bias_data, const cmsis_nn_dims output_dims, int8_t *output_data)
	Fast s8 version for 1x1 convolution (non-square shape) More...

int32_t	arm_convolve_1x1_s8_fast_get_buffer_size (const cmsis_nn_dims *input_dims)
	Get the required buffer size for arm_convolve_1x1_s8_fast. More...

arm_cmsis_nn_status	arm_convolve_fast_s16 (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int16_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int64_t bias_data, const cmsis_nn_dims output_dims, int16_t *output_data)
	Optimized s16 convolution function. More...

int32_t	arm_convolve_fast_s16_get_buffer_size (const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims)
	Get the required buffer size for fast s16 convolution function. More...

arm_cmsis_nn_status	arm_convolve_s16 (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int16_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int64_t bias_data, const cmsis_nn_dims output_dims, int16_t *output_data)
	Basic s16 convolution function. More...

int32_t	arm_convolve_s16_get_buffer_size (const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims)
	Get the required buffer size for s16 convolution function. More...

arm_cmsis_nn_status	arm_convolve_s8 (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int32_t bias_data, const cmsis_nn_dims output_dims, int8_t *output_data)
	Basic s8 convolution function. More...

int32_t	arm_convolve_s8_get_buffer_size (const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims)
	Get the required buffer size for s8 convolution function. More...

arm_cmsis_nn_status	arm_convolve_wrapper_s16 (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int16_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int64_t bias_data, const cmsis_nn_dims output_dims, int16_t *output_data)
	s16 convolution layer wrapper function with the main purpose to call the optimal kernel available in cmsis-nn to perform the convolution. More...

int32_t	arm_convolve_wrapper_s16_get_buffer_size (const cmsis_nn_conv_params conv_params, const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims, const cmsis_nn_dims output_dims)
	Get the required buffer size for arm_convolve_wrapper_s16. More...

arm_cmsis_nn_status	arm_convolve_wrapper_s8 (const cmsis_nn_context ctx, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input_data, const cmsis_nn_dims filter_dims, const int8_t filter_data, const cmsis_nn_dims bias_dims, const int32_t bias_data, const cmsis_nn_dims output_dims, int8_t *output_data)
	s8 convolution layer wrapper function with the main purpose to call the optimal kernel available in cmsis-nn to perform the convolution. More...

int32_t	arm_convolve_wrapper_s8_get_buffer_size (const cmsis_nn_conv_params conv_params, const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims, const cmsis_nn_dims output_dims)
	Get the required buffer size for arm_convolve_wrapper_s8. More...

arm_cmsis_nn_status	arm_depthwise_conv_3x3_s8 (const cmsis_nn_context ctx, const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input, const cmsis_nn_dims filter_dims, const int8_t kernel, const cmsis_nn_dims bias_dims, const int32_t bias, const cmsis_nn_dims output_dims, int8_t *output)
	Optimized s8 depthwise convolution function for 3x3 kernel size with some constraints on the input arguments(documented below). Refer arm_depthwise_conv_s8() for function argument details. More...

arm_cmsis_nn_status	arm_depthwise_conv_fast_s16 (const cmsis_nn_context ctx, const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int16_t input, const cmsis_nn_dims filter_dims, const int8_t kernel, const cmsis_nn_dims bias_dims, const int64_t bias, const cmsis_nn_dims output_dims, int16_t *output)
	Optimized s16 depthwise convolution function with constraint that in_channel equals out_channel. Refer arm_depthwise_conv_s16() for function argument details. More...

int32_t	arm_depthwise_conv_fast_s16_get_buffer_size (const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims)
	Get the required buffer size for optimized s16 depthwise convolution function with constraint that in_channel equals out_channel. More...

arm_cmsis_nn_status	arm_depthwise_conv_s16 (const cmsis_nn_context ctx, const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int16_t input, const cmsis_nn_dims filter_dims, const int8_t kernel, const cmsis_nn_dims bias_dims, const int64_t bias, const cmsis_nn_dims output_dims, int16_t *output)
	Basic s16 depthwise convolution function that doesn't have any constraints on the input dimensions. More...

	__attribute__ ((optimize("no-unroll-loops")))

arm_cmsis_nn_status	arm_depthwise_conv_s8 (const cmsis_nn_context ctx, const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input, const cmsis_nn_dims filter_dims, const int8_t kernel, const cmsis_nn_dims bias_dims, const int32_t bias, const cmsis_nn_dims output_dims, int8_t *output)
	Basic s8 depthwise convolution function that doesn't have any constraints on the input dimensions. More...

arm_cmsis_nn_status	arm_depthwise_conv_s8_opt (const cmsis_nn_context ctx, const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input, const cmsis_nn_dims filter_dims, const int8_t kernel, const cmsis_nn_dims bias_dims, const int32_t bias, const cmsis_nn_dims output_dims, int8_t *output)
	Optimized s8 depthwise convolution function with constraint that in_channel equals out_channel. Refer arm_depthwise_conv_s8() for function argument details. More...

int32_t	arm_depthwise_conv_s8_opt_get_buffer_size (const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims)
	Get the required buffer size for optimized s8 depthwise convolution function with constraint that in_channel equals out_channel. More...

arm_cmsis_nn_status	arm_depthwise_conv_wrapper_s16 (const cmsis_nn_context ctx, const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int16_t input, const cmsis_nn_dims filter_dims, const int8_t filter, const cmsis_nn_dims bias_dims, const int64_t bias, const cmsis_nn_dims output_dims, int16_t *output)
	Wrapper function to pick the right optimized s16 depthwise convolution function. More...

int32_t	arm_depthwise_conv_wrapper_s16_get_buffer_size (const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims, const cmsis_nn_dims output_dims)
	Get size of additional buffer required by arm_depthwise_conv_wrapper_s16() More...

arm_cmsis_nn_status	arm_depthwise_conv_wrapper_s8 (const cmsis_nn_context ctx, const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_per_channel_quant_params quant_params, const cmsis_nn_dims input_dims, const int8_t input, const cmsis_nn_dims filter_dims, const int8_t filter, const cmsis_nn_dims bias_dims, const int32_t bias, const cmsis_nn_dims output_dims, int8_t *output)
	Wrapper function to pick the right optimized s8 depthwise convolution function. More...

int32_t	arm_depthwise_conv_wrapper_s8_get_buffer_size (const cmsis_nn_dw_conv_params dw_conv_params, const cmsis_nn_dims input_dims, const cmsis_nn_dims filter_dims, const cmsis_nn_dims output_dims)
	Get size of additional buffer required by arm_depthwise_conv_wrapper_s8() More...

Description

Collection of convolution, depthwise convolution functions and their variants.

The convolution is implemented in 2 steps: im2col and General Matrix Multiplication(GEMM)

im2col is a process of converting each patch of image data into a column. After im2col, the convolution is computed as matrix-matrix multiplication.

To reduce the memory footprint, the im2col is performed partially. Each iteration, only a few column (i.e., patches) are generated followed by GEMM.

Function Documentation

◆ attribute()

__attribute__ ( (optimize("no-unroll-loops")) )

◆ arm_convolve_1_x_n_s8()

arm_cmsis_nn_status arm_convolve_1_x_n_s8	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. arm_convolve_1_x_n_s8_get_buffer_size will return the buffer_size if required The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). Range of conv_params->input_offset : [-127, 128] Range of conv_params->output_offset : [-128, 127]
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, 1, WK, C_IN] where WK is the horizontal spatial filter dimension
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Optional bias data pointer. Data type: int32
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int8

Returns: The function returns either ARM_CMSIS_NN_ARG_ERROR if argument constraints fail. or, ARM_CMSIS_NN_SUCCESS on successful completion.

Supported framework : TensorFlow Lite Micro
The following constrains on the arguments apply
1. input_dims->n equals 1
2. ouput_dims->w is a multiple of 4
3. Explicit constraints(since it is for 1xN convolution) -## input_dims->h equals 1 -## output_dims->h equals 1 -## filter_dims->h equals 1
  Todo:
  Remove constraint on output_dims->w to make the function generic.

◆ arm_convolve_1_x_n_s8_get_buffer_size()

int32_t arm_convolve_1_x_n_s8_get_buffer_size	(	const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims
	)

Parameters

[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, 1, WK, C_IN] where WK is the horizontal spatial filter dimension

Returns: The function returns required buffer size(bytes)

◆ arm_convolve_1x1_s8()

arm_cmsis_nn_status arm_convolve_1x1_s8	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. None is required by this function.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). Range of conv_params->input_offset : [-127, 128] Range of conv_params->output_offset : [-128, 127]
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, 1, 1, C_IN]
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Optional bias data pointer. Data type: int32
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int8

Returns: The function returns either ARM_CMSIS_NN_ARG_ERROR if argument constraints fail. or, ARM_CMSIS_NN_SUCCESS on successful completion.

Supported framework : TensorFlow Lite Micro
The following constrains on the arguments apply
1. conv_params->padding.w = conv_params->padding.h = 0

◆ arm_convolve_1x1_s8_fast()

arm_cmsis_nn_status arm_convolve_1x1_s8_fast	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. arm_convolve_1x1_s8_fast_get_buffer_size will return the buffer_size if required. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). Range of conv_params->input_offset : [-127, 128] Range of conv_params->output_offset : [-128, 127]
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, 1, 1, C_IN]
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Optional bias data pointer. Data type: int32
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int8

Returns: The function returns either ARM_CMSIS_NN_ARG_ERROR if argument constraints fail. or, ARM_CMSIS_NN_SUCCESS on successful completion.

Supported framework : TensorFlow Lite Micro
The following constrains on the arguments apply
1. input_dims->c is a multiple of 4
2. conv_params->padding.w = conv_params->padding.h = 0
3. conv_params->stride.w = conv_params->stride.h = 1

◆ arm_convolve_1x1_s8_fast_get_buffer_size()

int32_t arm_convolve_1x1_s8_fast_get_buffer_size ( const cmsis_nn_dims * input_dims )

Parameters

[in] input_dims Input (activation) dimensions

Returns: The function returns the required buffer size in bytes

◆ arm_convolve_fast_s16()

arm_cmsis_nn_status arm_convolve_fast_s16	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int16_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int64_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int16_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. arm_convolve_fast_s16_get_buffer_size will return the buffer_size if required. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). conv_params->input_offset : Not used conv_params->output_offset : Not used
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int16
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions. (filter_dims->w * filter_dims->h * input_dims->c) must not exceed 512
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Optional bias data pointer. Data type: int64
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int16

Returns: The function returns ARM_CMSIS_NN_SUCCESS

Supported framework: TensorFlow Lite micro
Additional memory is required for optimization. Refer to argument 'ctx' for details.
Implementation supports kernel volumes (filter width * filter height * input channels) < 512.

◆ arm_convolve_fast_s16_get_buffer_size()

int32_t arm_convolve_fast_s16_get_buffer_size	(	const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims
	)

Parameters

[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions

Returns: The function returns required buffer size(bytes)

◆ arm_convolve_s16()

arm_cmsis_nn_status arm_convolve_s16	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int16_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int64_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int16_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. arm_convolve_s16_get_buffer_size will return the buffer_size if required. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). conv_params->input_offset : Not used conv_params->output_offset : Not used
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int16
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Optional bias data pointer. Data type: int64
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int16

Returns: The function returns ARM_CMSIS_NN_SUCCESS

Supported framework: TensorFlow Lite micro
Additional memory is required for optimization. Refer to argument 'ctx' for details.

◆ arm_convolve_s16_get_buffer_size()

int32_t arm_convolve_s16_get_buffer_size	(	const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims
	)

Parameters

[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions

Returns: The function returns required buffer size(bytes)

◆ arm_convolve_s8()

arm_cmsis_nn_status arm_convolve_s8	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. arm_convolve_s8_get_buffer_size will return the buffer_size if required. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). Range of conv_params->input_offset : [-127, 128] Range of conv_params->output_offset : [-128, 127]
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Optional bias data pointer. Data type: int32
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int8

Returns: The function returns ARM_CMSIS_NN_SUCCESS

Supported framework: TensorFlow Lite micro
Additional memory is required for optimization. Refer to argument 'ctx' for details.

◆ arm_convolve_s8_get_buffer_size()

int32_t arm_convolve_s8_get_buffer_size	(	const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims
	)

Parameters

[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions

Returns: The function returns required buffer size(bytes)

◆ arm_convolve_wrapper_s16()

arm_cmsis_nn_status arm_convolve_wrapper_s16	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int16_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int64_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int16_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. arm_convolve_wrapper_s8_get_buffer_size will return the buffer_size if required The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). conv_params->input_offset : Not used conv_params->output_offset : Not used
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int16
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Bias data pointer. Data type: int64
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int16

Returns: The function returns either ARM_CMSIS_NN_ARG_ERROR if argument constraints fail. or, ARM_CMSIS_NN_SUCCESS on successful completion.

◆ arm_convolve_wrapper_s16_get_buffer_size()

int32_t arm_convolve_wrapper_s16_get_buffer_size	(	const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims,
		const cmsis_nn_dims *	output_dims
	)

Parameters

[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). conv_params->input_offset : Not used conv_params->output_offset : Not used
[in]	input_dims	Input (activation) dimensions. Format: [N, H, W, C_IN]
[in]	filter_dims	Filter dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]

Returns: The function returns required buffer size(bytes)

◆ arm_convolve_wrapper_s8()

arm_cmsis_nn_status arm_convolve_wrapper_s8	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Parameters

[in,out]	ctx	Function context that contains the additional buffer if required by the function. arm_convolve_wrapper_s8_get_buffer_size will return the buffer_size if required. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). Range of conv_params->input_offset : [-127, 128] Range of conv_params->output_offset : [-128, 127]
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN]
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Bias data pointer. Data type: int32
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[out]	output_data	Output data pointer. Data type: int8

Returns: The function returns either ARM_CMSIS_NN_ARG_ERROR if argument constraints fail. or, ARM_CMSIS_NN_SUCCESS on successful completion.

◆ arm_convolve_wrapper_s8_get_buffer_size()

int32_t arm_convolve_wrapper_s8_get_buffer_size	(	const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims,
		const cmsis_nn_dims *	output_dims
	)

Parameters

[in]	conv_params	Convolution parameters (e.g. strides, dilations, pads,...). Range of conv_params->input_offset : [-127, 128] Range of conv_params->output_offset : [-128, 127]
[in]	input_dims	Input (activation) dimensions. Format: [N, H, W, C_IN]
[in]	filter_dims	Filter dimensions. Format: [C_OUT, HK, WK, C_IN] where HK and WK are the spatial filter dimensions
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]

Returns: The function returns required buffer size(bytes)

◆ arm_depthwise_conv_3x3_s8()

arm_cmsis_nn_status arm_depthwise_conv_3x3_s8	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Returns

The function returns one of the following ARM_CMSIS_NN_ARG_ERROR - Unsupported dimension of tensors

Unsupported pad size along the x axis ARM_CMSIS_NN_SUCCESS - Successful operation

Supported framework : TensorFlow Lite Micro
The following constrains on the arguments apply
1. Number of input channel equals number of output channels
2. Filter height and width equals 3
3. Padding along x is either 0 or 1.

◆ arm_depthwise_conv_fast_s16()

arm_cmsis_nn_status arm_depthwise_conv_fast_s16	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int16_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int64_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int16_t *	output_data
	)

Returns: The function returns one of the following ARM_CMSIS_NN_ARG_ERROR - ctx-buff == NULL and arm_depthwise_conv_fast_s16_get_buffer_size() > 0 or input channel != output channel or ch_mult != 1

ARM_CMSIS_NN_SUCCESS - Successful operation

Supported framework: TensorFlow Lite
The following constrains on the arguments apply
1. Number of input channel equals number of output channels or ch_mult equals 1
Reccomended when number of channels is 4 or greater.

◆ arm_depthwise_conv_fast_s16_get_buffer_size()

int32_t arm_depthwise_conv_fast_s16_get_buffer_size	(	const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims
	)

Parameters

[in]	input_dims	Input (activation) tensor dimensions. Format: [1, H, W, C_IN] Batch argument N is not used.
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]

Returns: The function returns required buffer size in bytes

◆ arm_depthwise_conv_s16()

arm_cmsis_nn_status arm_depthwise_conv_s16	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int16_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int64_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int16_t *	output_data
	)

Parameters

[in,out]	ctx	Function context (e.g. temporary buffer). Check the function definition file to see if an additional buffer is required. Optional function {API}_get_buffer_size() provides the buffer size if an additional buffer is required. exists if additional memory is. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	dw_conv_params	Depthwise convolution parameters (e.g. strides, dilations, pads,...) conv_params->input_offset : Not used conv_params->output_offset : Not used
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN] Batch argument N is not used.
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Bias data pointer. Data type: int64
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[in,out]	output_data	Output data pointer. Data type: int16

Returns: The function returns ARM_CMSIS_NN_SUCCESS

Supported framework: TensorFlow Lite

◆ arm_depthwise_conv_s8()

arm_cmsis_nn_status arm_depthwise_conv_s8	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Parameters

[in,out]	ctx	Function context (e.g. temporary buffer). Check the function definition file to see if an additional buffer is required. Optional function {API}_get_buffer_size() provides the buffer size if an additional buffer is required exists if additional memory is. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	dw_conv_params	Depthwise convolution parameters (e.g. strides, dilations, pads,...) dw_conv_params->dilation is not used. Range of dw_conv_params->input_offset : [-127, 128] Range of dw_conv_params->input_offset : [-128, 127]
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [N, H, W, C_IN] Batch argument N is not used.
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Bias data pointer. Data type: int32
[in]	output_dims	Output tensor dimensions. Format: [N, H, W, C_OUT]
[in,out]	output_data	Output data pointer. Data type: int8

Returns: The function returns ARM_CMSIS_NN_SUCCESS

Supported framework: TensorFlow Lite

◆ arm_depthwise_conv_s8_opt()

arm_cmsis_nn_status arm_depthwise_conv_s8_opt	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Returns: The function returns one of the following ARM_CMSIS_NN_ARG_ERROR - input channel != output channel or ch_mult != 1 ARM_CMSIS_NN_SUCCESS - Successful operation

Note

If number of channels is not a multiple of 4, upto 3 elements outside the boundary will be read out for the following if MVE optimizations(Arm Helium Technology) are used.

Output shift
Output multiplier
Output bias
kernel

Supported framework: TensorFlow Lite
The following constrains on the arguments apply
1. Number of input channel equals number of output channels or ch_mult equals 1
Reccomended when number of channels is 4 or greater.

◆ arm_depthwise_conv_s8_opt_get_buffer_size()

int32_t arm_depthwise_conv_s8_opt_get_buffer_size	(	const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims
	)

Parameters

[in]	input_dims	Input (activation) tensor dimensions. Format: [1, H, W, C_IN] Batch argument N is not used.
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]

Returns: The function returns required buffer size in bytes

◆ arm_depthwise_conv_wrapper_s16()

arm_cmsis_nn_status arm_depthwise_conv_wrapper_s16	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int16_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int64_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int16_t *	output_data
	)

Parameters

[in,out]	ctx	Function context (e.g. temporary buffer). Check the function definition file to see if an additional buffer is required. Optional function {API}_get_buffer_size() provides the buffer size if required. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	dw_conv_params	Depthwise convolution parameters (e.g. strides, dilations, pads,...) dw_conv_params->dilation is not used. Range of dw_conv_params->input_offset : Not used Range of dw_conv_params->output_offset : Not used
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [H, W, C_IN] Batch argument N is not used and assumed to be 1.
[in]	input_data	Input (activation) data pointer. Data type: int16
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Bias data pointer. Data type: int64
[in]	output_dims	Output tensor dimensions. Format: [1, H, W, C_OUT]
[in,out]	output_data	Output data pointer. Data type: int16

Returns: The function returns ARM_CMSIS_NN_SUCCESS - Successful completion.

Supported framework: TensorFlow Lite
Picks one of the the following functions
1. arm_depthwise_conv_s16()
2. arm_depthwise_conv_fast_s16() - Cortex-M CPUs with DSP extension only

◆ arm_depthwise_conv_wrapper_s16_get_buffer_size()

int32_t arm_depthwise_conv_wrapper_s16_get_buffer_size	(	const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims,
		const cmsis_nn_dims *	output_dims
	)

Parameters

[in]	dw_conv_params	Depthwise convolution parameters (e.g. strides, dilations, pads,...) Range of dw_conv_params->input_offset : Not used Range of dw_conv_params->input_offset : Not used
[in]	input_dims	Input (activation) tensor dimensions. Format: [H, W, C_IN] Batch argument N is not used and assumed to be 1.
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]
[in]	output_dims	Output tensor dimensions. Format: [1, H, W, C_OUT]

Returns: Size of additional memory required for optimizations in bytes.

◆ arm_depthwise_conv_wrapper_s8()

arm_cmsis_nn_status arm_depthwise_conv_wrapper_s8	(	const cmsis_nn_context *	ctx,
		const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const cmsis_nn_dims *	input_dims,
		const int8_t *	input_data,
		const cmsis_nn_dims *	filter_dims,
		const int8_t *	filter_data,
		const cmsis_nn_dims *	bias_dims,
		const int32_t *	bias_data,
		const cmsis_nn_dims *	output_dims,
		int8_t *	output_data
	)

Parameters

[in,out]	ctx	Function context (e.g. temporary buffer). Check the function definition file to see if an additional buffer is required. Optional function {API}_get_buffer_size() provides the buffer size if required. The caller is expected to clear the buffer ,if applicable, for security reasons.
[in]	dw_conv_params	Depthwise convolution parameters (e.g. strides, dilations, pads,...) dw_conv_params->dilation is not used. Range of dw_conv_params->input_offset : [-127, 128] Range of dw_conv_params->output_offset : [-128, 127]
[in]	quant_params	Per-channel quantization info. It contains the multiplier and shift values to be applied to each output channel
[in]	input_dims	Input (activation) tensor dimensions. Format: [H, W, C_IN] Batch argument N is not used and assumed to be 1.
[in]	input_data	Input (activation) data pointer. Data type: int8
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]
[in]	filter_data	Filter data pointer. Data type: int8
[in]	bias_dims	Bias tensor dimensions. Format: [C_OUT]
[in]	bias_data	Bias data pointer. Data type: int32
[in]	output_dims	Output tensor dimensions. Format: [1, H, W, C_OUT]
[in,out]	output_data	Output data pointer. Data type: int8

Returns: The function returns ARM_CMSIS_NN_SUCCESS - Successful completion.

Supported framework: TensorFlow Lite
Picks one of the the following functions
1. arm_depthwise_conv_s8()
2. arm_depthwise_conv_3x3_s8() - Cortex-M CPUs with DSP extension only
3. arm_depthwise_conv_s8_opt()
Check details of arm_depthwise_conv_s8_opt() for potential data that can be accessed outside of the boundary.

◆ arm_depthwise_conv_wrapper_s8_get_buffer_size()

int32_t arm_depthwise_conv_wrapper_s8_get_buffer_size	(	const cmsis_nn_dw_conv_params *	dw_conv_params,
		const cmsis_nn_dims *	input_dims,
		const cmsis_nn_dims *	filter_dims,
		const cmsis_nn_dims *	output_dims
	)

Parameters

[in]	dw_conv_params	Depthwise convolution parameters (e.g. strides, dilations, pads,...) Range of dw_conv_params->input_offset : [-127, 128] Range of dw_conv_params->input_offset : [-128, 127]
[in]	input_dims	Input (activation) tensor dimensions. Format: [H, W, C_IN] Batch argument N is not used and assumed to be 1.
[in]	filter_dims	Filter tensor dimensions. Format: [1, H, W, C_OUT]
[in]	output_dims	Output tensor dimensions. Format: [1, H, W, C_OUT]

Returns: Size of additional memory required for optimizations in bytes.

Functions

Description

Function Documentation

◆ __attribute__()

◆ arm_convolve_1_x_n_s8()

◆ arm_convolve_1_x_n_s8_get_buffer_size()

◆ arm_convolve_1x1_s8()

◆ arm_convolve_1x1_s8_fast()

◆ arm_convolve_1x1_s8_fast_get_buffer_size()

◆ arm_convolve_fast_s16()

◆ arm_convolve_fast_s16_get_buffer_size()

◆ arm_convolve_s16()

◆ arm_convolve_s16_get_buffer_size()

◆ arm_convolve_s8()

◆ arm_convolve_s8_get_buffer_size()

◆ arm_convolve_wrapper_s16()

◆ arm_convolve_wrapper_s16_get_buffer_size()

◆ arm_convolve_wrapper_s8()

◆ arm_convolve_wrapper_s8_get_buffer_size()

◆ arm_depthwise_conv_3x3_s8()

◆ arm_depthwise_conv_fast_s16()

◆ arm_depthwise_conv_fast_s16_get_buffer_size()

◆ arm_depthwise_conv_s16()

◆ arm_depthwise_conv_s8()

◆ arm_depthwise_conv_s8_opt()

◆ arm_depthwise_conv_s8_opt_get_buffer_size()

◆ arm_depthwise_conv_wrapper_s16()

◆ arm_depthwise_conv_wrapper_s16_get_buffer_size()

◆ arm_depthwise_conv_wrapper_s8()

◆ arm_depthwise_conv_wrapper_s8_get_buffer_size()

◆ attribute()