Parameter Extraction of PSP MOSFET Model in Multi-core Zynq SoC Platform

Abstract

Abstract Zynq System on Chip (SoC) is widely used architecture to accelerate the computational intensive application with its ARM Processing System (PS) and FPGA fabric - Programmable Logic (PL). In this paper, parameter extraction of PSP MOSFET model is studied as a computational intensive process and its function is accelerated by HW/SW co-design on Zynq SoC using its dual-core Asymmetric Multiprocessing (AMP) architecture. Particle Swarm Optimization (PSO) algorithm is being applied to extract PSP model parameters in order to accurately match the model generated response with measured characteristic. This paper presents the acceleration model for two major processes of parameter extraction: 1) Optimization algorithm (PSO), and 2) Model library (SiMKit) execution. First, the PSO algorithm is implemented as hardware on PL. Second, PSP MOSFET library SiMKit is implemented as a software on PS. Implementation results show 73.06% improvement in PSO execution time compared to ARM-based software implementation. Dual-core AMP configuration results into 28.54% better performance in SiMKit execution time compared to a single-core processor.