4G Turbo-V Intel® FPGA IP User Guide

ID 683882
Date 4/01/2024
Public
Document Table of Contents

4.1.4. Latency Calculations

The overall latency of the downlink accelerator is the sum of the latency of the subcomponents, CRC, encoder, subblock-interleaver, pruning, and the output latency L(output). The calculations do not include control and data overhead delay, which may vary from block to block. Therefore, the calculated D is the shortest processing delay.

L(crc) = K/8+8

L(encoder) = K/8+14

L(subblock-interleaver) = Kπ/8+42

L(pruning) = Kπ/8+14

L(output) = ceil(E/24)+46

Example 1: K = 6144, Kπ = 6176, E = 18444

L(crc) = 6144/8+8 = 776 cycles

L(encoder) = 6144/8+14 = 782 cycles

L(subblock-interleaver) = 6176/8+42 = 814 cycles

L(pruning) = 6176/8+14 = 786 cycles

L(output) = ceil(18444/24)+46 = 815 cycles

L(downlink) = 776 + 782 + 814 + 786 + 815 = 3973 cycles

Example 2: K = 40, Kπ = 64, E = 132

L(crc) = 40/8+8 = 13 cycles

L(encoder) = 40/8+14 = 19 cycles

L(subblock-interleaver) = 64/8+42 = 50 cycles

L(pruning) = 64/8+14 = 22 cycles

L(output) = ceil(132/24)+46 = 52 cycles

L(downlink) = 13 + 19 + 50 + 22 + 52 = 156 cycles