面向机载应用的领域专用加速器研究

机载环境下数据处理规模的剧增以及人机混合智能的应用使得传统的以CPU 为核心计算单元的架构 已不能满足计算需求。在满足延时、精度等指标的情况下，选用高能效的处理器或处理器组合来快速准确地处 理这些数据成为机载计算领域面临的重要问题。按照常规处理器、领域专用加速器两大类型对各自主要代表性 处理器的架构特点进行了分析和总结，得出了各类处理器在机载情况下的主要适用场景和应用情况。根据领域 专用的设计思想开发了面向数据关联应用的专用加速器，对数据关联算法中的统计距离计算和分簇处理这两个 计算瓶颈进行了定制化的加速设计，并在基于FPGA 的平台上进行了测试验证，结果表明，加速器对于统计距 离计算的加速效果约为FT2000/4 单核性能的10 倍，对于分簇处理的加速效果约为FT2000/4 单核性能的3 倍， 整体运算速度相比FT2000/4 处理器的单核提升了5 倍。

Research on Domain Specific Accelerators for Airborne Applications

The rapid increase in the scale of data processing and the application of human-computer hybrid intelligence have made the traditional architecture with CPU as the main computing unit no longer able to meet the computing needs in the airborne environment. In order to satisfy indicators such as delay and accuracy, selecting a high energy efficiency processor or combination of processors to process these data quickly and efficiently has become an important issue in the field of airborne computing. The architecture features, applicable scenarios, and application conditions of the respective main representative processors are analyzed and summarized, according to the general-purpose and special-purpose types. A dedicated accelerator for data-related applications has been designed and implemented on an FPGA-based platform. The results show that the acceleration effect of the accelerator for statistical distance calculation is about 10 times that of FT2000/4 single-core performance, and the acceleration effect for clustering processing is about 3 times that of FT2000/4 single-core performance. The overall computing speed is increased by 5times.