Functions
arm_cmsis_nn_status	arm_nn_depthwise_conv_nt_t_padded_s8 (const int8_t lhs, const int8_t rhs, const int32_t input_offset, const int32_t active_ch, const int32_t total_ch, const int32_t out_shift, const int32_t out_mult, const int32_t out_offset, const int32_t activation_min, const int32_t activation_max, const uint16_t row_x_col, const int32_t const output_bias, int8_t out)
	Depthwise convolution of transposed rhs matrix with 4 lhs matrices. To be used in padded cases where the padding is -lhs_offset(Range: int8). Dimensions are the same for lhs and rhs. More...

int16_t *	arm_nn_depthwise_conv_nt_t_s16 (const int16_t lhs, const int8_t rhs, const uint16_t num_ch, const int32_t out_shift, const int32_t out_mult, const int32_t activation_min, const int32_t activation_max, const uint16_t row_x_col, const int64_t const output_bias, int16_t out)
	Depthwise convolution of transposed rhs matrix with 4 lhs matrices. To be used in non-padded cases. Dimensions are the same for lhs and rhs. More...

arm_cmsis_nn_status	arm_nn_depthwise_conv_nt_t_s8 (const int8_t lhs, const int8_t rhs, const int32_t input_offset, const int32_t active_ch, const int32_t total_ch, const int32_t out_shift, const int32_t out_mult, const int32_t out_offset, const int32_t activation_min, const int32_t activation_max, const uint16_t row_x_col, const int32_t const output_bias, int8_t out)
	Depthwise convolution of transposed rhs matrix with 4 lhs matrices. To be used in non-padded cases. Dimensions are the same for lhs and rhs. More...

arm_cmsis_nn_status	arm_nn_mat_mul_core_1x_s8 (int32_t row_elements, const int32_t skipped_row_elements, const int8_t row_base_ref, const int8_t col_base_ref, const int32_t out_ch, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const int32_t bias, int8_t output)
	General Vector by Matrix multiplication with requantization and storage of result. More...

int8_t *	arm_nn_mat_mul_core_4x_s8 (const int32_t row_elements, const int32_t offset, const int8_t row_base, const int8_t col_base_ref, const int32_t out_ch, const cmsis_nn_conv_params conv_params, const cmsis_nn_per_channel_quant_params quant_params, const int32_t bias, int8_t output)
	Matrix-multiplication with requantization & activation function for four rows and one column. More...

int16_t *	arm_nn_mat_mult_kernel_s16 (const int8_t input_a, const int16_t input_b, const int32_t output_ch, const int32_t out_shift, const int32_t out_mult, const int16_t activation_min, const int16_t activation_max, const int32_t num_col_a, const int64_t const output_bias, int16_t out_0)
	Matrix-multiplication function for convolution with per-channel requantization for 16 bits convolution. More...

arm_cmsis_nn_status	arm_nn_mat_mult_nt_t_s8 (const int8_t lhs, const int8_t rhs, const int32_t bias, int8_t dst, const int32_t dst_multipliers, const int32_t dst_shifts, const int32_t lhs_rows, const int32_t rhs_rows, const int32_t rhs_cols, const int32_t lhs_offset, const int32_t dst_offset, const int32_t activation_min, const int32_t activation_max, const int32_t rhs_cols_offset)
	General Matrix-multiplication function with per-channel requantization. This function assumes: More...

Description

Support functions for Convolution and DW Convolution

Function Documentation

◆ arm_nn_depthwise_conv_nt_t_padded_s8()

arm_cmsis_nn_status arm_nn_depthwise_conv_nt_t_padded_s8	(	const int8_t *	lhs,
		const int8_t *	rhs,
		const int32_t	lhs_offset,
		const int32_t	active_ch,
		const int32_t	total_ch,
		const int32_t *	out_shift,
		const int32_t *	out_mult,
		const int32_t	out_offset,
		const int32_t	activation_min,
		const int32_t	activation_max,
		const uint16_t	row_x_col,
		const int32_t *const	output_bias,
		int8_t *	out
	)

Parameters

[in]	lhs	Input left-hand side matrix
[in]	rhs	Input right-hand side matrix (transposed)
[in]	lhs_offset	LHS matrix offset(input offset). Range: -127 to 128
[in]	active_ch	Subset of total_ch processed
[in]	total_ch	Number of channels in LHS/RHS
[in]	out_shift	Per channel output shift. Length of vector is equal to number of channels
[in]	out_mult	Per channel output multiplier. Length of vector is equal to number of channels
[in]	out_offset	Offset to be added to the output values. Range: -127 to 128
[in]	activation_min	Minimum value to clamp the output to. Range: int8
[in]	activation_max	Maximum value to clamp the output to. Range: int8
[in]	row_x_col	(row_dimension * col_dimension) of LHS/RHS matrix
[in]	output_bias	Per channel output bias. Length of vector is equal to number of channels
[in]	out	Output pointer

Returns

The function returns one of the two

Updated output pointer if an implementation is available
NULL if no implementation is available.

Note

If number of channels is not a multiple of 4, upto 3 elements outside the boundary will be read out for the following.

Output shift
Output multiplier
Output bias
rhs

◆ arm_nn_depthwise_conv_nt_t_s16()

int16_t * arm_nn_depthwise_conv_nt_t_s16	(	const int16_t *	lhs,
		const int8_t *	rhs,
		const uint16_t	num_ch,
		const int32_t *	out_shift,
		const int32_t *	out_mult,
		const int32_t	activation_min,
		const int32_t	activation_max,
		const uint16_t	row_x_col,
		const int64_t *const	output_bias,
		int16_t *	out
	)

Parameters

[in]	lhs	Input left-hand side matrix
[in]	rhs	Input right-hand side matrix (transposed)
[in]	num_ch	Number of channels in LHS/RHS
[in]	out_shift	Per channel output shift. Length of vector is equal to number of channels.
[in]	out_mult	Per channel output multiplier. Length of vector is equal to number of channels.
[in]	activation_min	Minimum value to clamp the output to. Range: int8
[in]	activation_max	Maximum value to clamp the output to. Range: int8
[in]	row_x_col	(row_dimension * col_dimension) of LHS/RHS matrix
[in]	output_bias	Per channel output bias. Length of vector is equal to number of channels.
[in]	out	Output pointer

Returns

The function returns one of the two

Updated output pointer if an implementation is available
NULL if no implementation is available.

Note

If number of channels is not a multiple of 4, upto 3 elements outside the boundary will be read out for the following.

Output shift
Output multiplier
Output bias
rhs

◆ arm_nn_depthwise_conv_nt_t_s8()

arm_cmsis_nn_status arm_nn_depthwise_conv_nt_t_s8	(	const int8_t *	lhs,
		const int8_t *	rhs,
		const int32_t	lhs_offset,
		const int32_t	active_ch,
		const int32_t	total_ch,
		const int32_t *	out_shift,
		const int32_t *	out_mult,
		const int32_t	out_offset,
		const int32_t	activation_min,
		const int32_t	activation_max,
		const uint16_t	row_x_col,
		const int32_t *const	output_bias,
		int8_t *	out
	)

Parameters

[in]	lhs	Input left-hand side matrix
[in]	rhs	Input right-hand side matrix (transposed)
[in]	lhs_offset	LHS matrix offset(input offset). Range: -127 to 128
[in]	active_ch	Subset of total_ch processed
[in]	total_ch	Number of channels in LHS/RHS
[in]	out_shift	Per channel output shift. Length of vector is equal to number of channels.
[in]	out_mult	Per channel output multiplier. Length of vector is equal to number of channels.
[in]	out_offset	Offset to be added to the output values. Range: -127 to 128
[in]	activation_min	Minimum value to clamp the output to. Range: int8
[in]	activation_max	Maximum value to clamp the output to. Range: int8
[in]	row_x_col	(row_dimension * col_dimension) of LHS/RHS matrix
[in]	output_bias	Per channel output bias. Length of vector is equal to number of channels.
[in]	out	Output pointer

Returns

The function returns one of the two

Updated output pointer if an implementation is available
NULL if no implementation is available.

Note

If number of channels is not a multiple of 4, upto 3 elements outside the boundary will be read out for the following.

Output shift
Output multiplier
Output bias
rhs

◆ arm_nn_mat_mul_core_1x_s8()

arm_cmsis_nn_status arm_nn_mat_mul_core_1x_s8	(	int32_t	row_elements,
		const int32_t	skipped_row_elements,
		const int8_t *	row_base_ref,
		const int8_t *	col_base_ref,
		const int32_t	out_ch,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const int32_t *	bias,
		int8_t *	output
	)

Parameters

[in]	row_elements	number of row elements
[in]	skipped_row_elements	number of row elements skipped due to padding. row_elements + skipped_row_elements = (kernel_x * kernel_y) * input_ch
[in]	row_base_ref	pointer to row operand
[in]	col_base_ref	pointer to col operand
[out]	out_ch	Number of output channels
[in]	conv_params	Pointer to convolution parameters like offsets and activation values
[in]	quant_params	Pointer to per-channel quantization parameters
[in]	bias	Pointer to optional per-channel bias
[out]	output	Pointer to output where int8 results are stored.

Returns: The function performs matrix(row_base_ref) multiplication with vector(col_base_ref) and scaled result is stored in memory.

Pseudo-code *output = 0 sum_col = 0 for (j = 0; j < out_ch; j++) for (i = 0; i < row_elements; i++) *output += row_base_ref[i] * col_base_ref[i] sum_col += col_base_ref[i] scale sum_col using quant_params and bias store result in 'output'

◆ arm_nn_mat_mul_core_4x_s8()

int8_t * arm_nn_mat_mul_core_4x_s8	(	const int32_t	row_elements,
		const int32_t	offset,
		const int8_t *	row_base,
		const int8_t *	col_base,
		const int32_t	out_ch,
		const cmsis_nn_conv_params *	conv_params,
		const cmsis_nn_per_channel_quant_params *	quant_params,
		const int32_t *	bias,
		int8_t *	output
	)

Parameters

[in]	row_elements	number of row elements
[in]	offset	offset between rows. Can be the same as row_elements. For e.g, in a 1x1 conv scenario with stride as 1.
[in]	row_base	pointer to row operand
[in]	col_base	pointer to col operand
[in]	out_ch	Number of output channels
[in]	conv_params	Pointer to convolution parameters like offsets and activation values
[in]	quant_params	Pointer to per-channel quantization parameters
[in]	bias	Pointer to per-channel bias
[out]	output	Pointer to output where int8 results are stored.

Returns: The function returns the updated output pointer or NULL if implementation is not available.

Compliant to TFLM int8 specification. MVE implementation only

◆ arm_nn_mat_mult_kernel_s16()

int16_t * arm_nn_mat_mult_kernel_s16	(	const int8_t *	input_a,
		const int16_t *	input_b,
		const int32_t	output_ch,
		const int32_t *	out_shift,
		const int32_t *	out_mult,
		const int16_t	activation_min,
		const int16_t	activation_max,
		const int32_t	num_col_a,
		const int64_t *const	output_bias,
		int16_t *	out_0
	)

Parameters

[in]	input_a	pointer to operand A
[in]	input_b	pointer to operand B, always consists of 2 vectors.
[in]	output_ch	number of rows of A
[in]	out_shift	pointer to per output channel requantization shift parameter.
[in]	out_mult	pointer to per output channel requantization multiplier parameter.
[in]	activation_min	minimum value to clamp the output to. Range : int16
[in]	activation_max	maximum value to clamp the output to. Range : int16
[in]	num_col_a	number of columns of A
[in]	output_bias	per output channel bias. Range : int64
[in,out]	out_0	pointer to output

Returns

The function returns one of the two

The incremented output pointer for a successful operation or
NULL if implementation is not available.

This function does the matrix multiplication of weight matrix for all output channels with 2 columns from im2col and produces two elements/output_channel. The outputs are clamped in the range provided by activation min and max. Supported framework: TensorFlow Lite micro.

◆ arm_nn_mat_mult_nt_t_s8()

arm_cmsis_nn_status arm_nn_mat_mult_nt_t_s8	(	const int8_t *	lhs,
		const int8_t *	rhs,
		const int32_t *	bias,
		int8_t *	dst,
		const int32_t *	dst_multipliers,
		const int32_t *	dst_shifts,
		const int32_t	lhs_rows,
		const int32_t	rhs_rows,
		const int32_t	rhs_cols,
		const int32_t	lhs_offset,
		const int32_t	dst_offset,
		const int32_t	activation_min,
		const int32_t	activation_max,
		const int32_t	rhs_cols_offset
	)

LHS input matrix NOT transposed (nt)
RHS input matrix transposed (t)

Note: This operation also performs the broadcast bias addition before the requantization

Parameters

[in]	lhs	Pointer to the LHS input matrix
[in]	rhs	Pointer to the RHS input matrix
[in]	bias	Pointer to the bias vector. The length of this vector is equal to the number of output columns (or RHS input rows)
[out]	dst	Pointer to the output matrix with "m" rows and "n" columns
[in]	dst_multipliers	Pointer to the multipliers vector needed for the per-channel requantization. The length of this vector is equal to the number of output columns (or RHS input rows)
[in]	dst_shifts	Pointer to the shifts vector needed for the per-channel requantization. The length of this vector is equal to the number of output columns (or RHS input rows)
[in]	lhs_rows	Number of LHS input rows
[in]	rhs_rows	Number of RHS input rows
[in]	rhs_cols	Number of LHS/RHS input columns
[in]	lhs_offset	Offset to be applied to the LHS input value
[in]	dst_offset	Offset to be applied the output result
[in]	activation_min	Minimum value to clamp down the output. Range : int8
[in]	activation_max	Maximum value to clamp up the output. Range : int8
[in]	rhs_cols_offset	Offset between input columns. Used to handle non-unity strides Expected value : x * rhs_cols, where x >= 1

Returns: The function returns ARM_CMSIS_NN_SUCCESS

Functions

Description

Function Documentation

◆ arm_nn_depthwise_conv_nt_t_padded_s8()

◆ arm_nn_depthwise_conv_nt_t_s16()

◆ arm_nn_depthwise_conv_nt_t_s8()

◆ arm_nn_mat_mul_core_1x_s8()

◆ arm_nn_mat_mul_core_4x_s8()

◆ arm_nn_mat_mult_kernel_s16()

◆ arm_nn_mat_mult_nt_t_s8()