Abstract：Timely and accurate system processing ability of missile electronic system against high-speed weak and small targets is one of the key factors for the missile to lock and successful strike the target. The computing resources are the core resources of target information processing. A hybrid architecture of target information intelligent processing based on heterogeneous computing resources of GPU and FPGA is proposed, which supports neural network algorithm based target information processing and software development by jointly utilizing GPU parallel processing and AI coprocessing ability and FPGA programmable extension ability. The standardization and generality design of proposed architecture meets the needs of on-demand customization, functional expansion and flexible configuration. The test results show that the proposed system can meet the requirements of in-time processing and agility of missile electronic system.
何健, 袁双, 李芳, 杨贝尔, 李庆波, 何志敏. 基于GPU+FPGA异构计算资源的目标信息智能处理技术[J]. 空天防御, 2021, 4(3): 10-16.
HE Jian, YUAN Shuang, LI Fang, YANG Beier, LI Qingbo, HE Zhimin. Target Information Intelligent Processing Technology Based on GPU+ FPGA Heterogeneous Computing Resources. Air & Space Defense, 2021, 4(3): 10-16.
