Ефективність оптимізації коду в середовищі високорівневого синтезу на платформі SOC ZYNQ-7000 у порівнянні з сучасними процесорними архітектурами

Abstract

The thesis investigates the efficiency of code optimization on the SoC Zynq7000 platform in comparison with modern CPU architectures such as AMD Ryzen 9, Apple M1 and Raspberry Pi. Experiments were carried out for matrix multiplication (classical and Winograd), fast Fourier transform and wavelet transforms using Haar and Daubechies wavelets. Several optimization models were analyzed, including memory access optimization, loop unrolling, function inlining and their combinations. The results demonstrate a significant reduction of execution time. The obtained recommendations can be used to design highperformance SoC-based systems for signal processing and other computationally loaded algorithms