In this paper, a novel approach to accelerate parameter extraction process of surface-potential-based (PSP) MOSFET model is presented for the submicron MOS transistor. To reduce the computational time, Field Programmable Gate Array (FPGA) implementation of PSP model library — SiMKit is demonstrated using Xilinx’s Zynq Ultrascale+MPSoC (Multi-processor System-on-Chip) platform. Parameter extraction is carried out using Particle Swarm Optimization (PSO) algorithm for 65nm technology nMOS devices. With the available measurement data, 32 various PSP model parameters are extracted. Experimental results validate the performance and accuracy of parameter extraction by achieving Root Mean Square Error below 10% for various current–voltage characteristics of nMOS device. 41.57% acceleration in execution time for extraction process is achieved by Zynq Ultrascale+MPSoC platform compared to the conventional computer-based software approach. In addition, various design optimization directives are explored, and their performances are compared as a part of RTL generation of SiMKit.

中文翻译：

---

加快SoC平台上PSP MOSFET模型的参数提取

在本文中，提出了一种用于加速亚微米 MOS 晶体管的基于表面电位 (PSP) MOSFET 模型的参数提取过程的新方法。为了减少计算时间，PSP 模型库的现场可编程门阵列 (FPGA) 实施 - SiMKit 使用 Xilinx 的 Zynq 进行了演示超大规模+MPSoC（多处理器片上系统）平台。使用粒子群优化 (PSO) 算法对 65 进行参数提取纳米技术 nMOS 器件。利用可用的测量数据，提取了 32 个不同的 PSP 模型参数。实验结果通过使 nMOS 器件的各种电流-电压特性的均方根误差低于 10%，验证了参数提取的性能和准确性。Zynq 将提取过程的执行时间加快了 41.57%超大规模+MPSoC平台与传统的基于计算机的软件方法相比。此外，还探索了各种设计优化指令，并将它们的性能作为 SiMKit 的 RTL 生成的一部分进行了比较。