kernel name: kfun_mult_cascade - 2023.2 English

This kernel is used to compute the final result. It accumulates the results of each transducer.

• Template params:

  • T: type of the operation;
  • NUM_LINE_t: number of scanlines;
  • NUM_ELEMENT_t: number of transducer’s elements;
  • NUM_SAMPLE_t: number of time samples;
  • NUM_SEG_t: number of segments in one scanline;
  • NUM_DEP_SEG_t: number of data in one segment;
  • VECDIM_mult_t: dimension of the SIMD to be performed. Addressed in the Xilinx UG1076, it depends on the type chosen;
  • LEN_IN_mult_t: number of inputdata per invoking;
  • LEN_OUT_mult_t: number of outputdata per invoking;
  • LEN32b_PARA_mult_t: number of inputdata per invoking;
  • MULT_ID_t: default ID of mult kernel;

• kernel Inputs:

  • para_const: kernel_mult’s kernel kfun_mult_0 self-used structural parameter include iter_line, iter_element to record the number of lines processed;
  • local_data_0: kernel_mult’s kernel kfun_mult_0 self-used local array, used to store accumulated temporary results;
  • local_data_1: kernel_mult’s kernel kfun_mult_0 self-used local array, used to store accumulated temporary results. To avoid bank conflict, this data memory is applied;
  • p_in_cascade: Results of kfun_mult_pre.

• kernel Outputs:

  • p_out_cascade: Output to next mult kernel;
  • p_out_mult: Output one segment of final result;