CMSIS-NN  Version 3.1.0
CMSIS NN Software Library
 All Data Structures Files Functions Variables Enumerations Enumerator Macros Groups Pages
arm_nn_mat_mult_s8.c File Reference


q7_t * arm_nn_mat_mult_s8 (const q7_t *input_row, const q7_t *input_col, const uint16_t output_ch, const uint16_t col_batches, const int32_t *output_shift, const int32_t *output_mult, const int32_t out_offset, const int32_t col_offset, const int32_t row_offset, const int16_t activation_min, const int16_t activation_max, const uint16_t row_len, const int32_t *const bias, q7_t *out)
 General Matrix-multiplication function with per-channel requantization. More...

Function Documentation

q7_t* arm_nn_mat_mult_s8 ( const q7_t *  input_row,
const q7_t *  input_col,
const uint16_t  output_ch,
const uint16_t  col_batches,
const int32_t *  output_shift,
const int32_t *  output_mult,
const int32_t  out_offset,
const int32_t  col_offset,
const int32_t  row_offset,
const int16_t  out_activation_min,
const int16_t  out_activation_max,
const uint16_t  row_len,
const int32_t *const  bias,
q7_t *  out 
[in]input_rowpointer to row operand
[in]input_colpointer to col operand
[in]output_chnumber of rows of input_row
[in]col_batchesnumber of column batches. Range: 1 to 4
[in]output_shiftpointer to per output channel requantization shift parameter.
[in]output_multpointer to per output channel requantization multiplier parameter.
[in]out_offsetoutput tensor offset.
[in]col_offsetinput tensor(col) offset.
[in]row_offsetkernel offset(row). Not used.
[in]out_activation_minminimum value to clamp the output to. Range : int8
[in]out_activation_maxmaximum value to clamp the output to. Range : int8
[in]row_lennumber of elements in each row
[in]biasper output channel bias. Range : int32
[in,out]outpointer to output
The function returns one of the two
  1. The incremented output pointer for a successful operation or
  2. NULL if implementation is not available.

Supported framework: TensorFlow Lite

References arm_nn_requantize(), MAX, and MIN.

Referenced by arm_convolve_s8().