计算机技术应用研究室知识库

采用GPU和最新的CUDA并行架构设计了射电天文多相滤波器组,并对其性能指标进行了测试和分析。利用GPU强大的浮点数计算和高效并行执行能力实现了PFB、FFT算法加速,改善了多相滤波器组算法的执行效率。实验结果表明,设计的多相滤波器组具有一定的灵活性和可扩展性,能够实现射电信号的高速滤波及信道化,可有效提高射电望远镜数字终端算法的并行数据处理能力和计算效率。

Polyphase filter banks was designed with the GPU and the latest NVIDIA CUDA parallel architecture for radio astronomy, as well as its performance was tested and analyzed. The using of GPU's powerful floating-point calculations and high performance parallel execution capabilities accelerated PFB and FFT algorithms, and improved the filter banksefficiency. Experiment results show us, thepolyphase filter banks was designed in this paper has a certain flexibility and scalability, implement high-speed filtering and high-speed channelization, improved the computing efficiency and parallel data processing capabilities of astronomical digital backend algorithm.