摘要
GPU具有大规模并行运算的特点,使用GPU作为计算节点可以大量节省硬件投入成本。论文通过对GPU硬件架构分析研究GPU做为新一代信号处理平台的优势,通过对CUDA软件架构的分析,研究影响信号处理算法在GPU上的性能的主要因素和提升性能的方法。最后在Jetson agx xavier平台上实现波束形成算法,并和CPU与DSP上的运算性能进行对比。
GPU has the characteristics of large-scale parallel computing.Using GPU as computing node can save a lot of hard⁃ware investment cost.This paper studies the advantages of GPU as a new generation of signal processing platform through the analy⁃sis of GPU hardware architecture,and studies the main factors affecting the performance of signal processing algorithms on GPU and methods to improve performance through the analysis of CUDA software architecture.Finally,the beamforming algorithm is imple⁃mented on the Jetson agx xavier platform and compared with the computing performance on the CPU and DSP.
作者
刘建涛
张海彬
LIU Jiantao;ZHANG Haibin(The Third Military Representative Office of Shenyang Bureau of Naval Equipment Department in Shenyang District,Shenyang 110003;Shenyang Liaohai Equipment Co.,Ltd.,Shenyang 110003)
出处
《舰船电子工程》
2022年第1期146-149,共4页
Ship Electronic Engineering
