中国科学技术大学学报 ›› 2015, Vol. 45 ›› Issue (7): 608-613.DOI: 10.3969/j.issn.0253-2778.2015.07.012

• 科研论文 • 上一篇    

面向高数据并行架构的原位FFT算法

  

  1. 1.合肥工业大学计算机与信息学院,安徽合肥, 230009; 2.中国电子科技集团公司第三十八研究所,安徽合肥,230088; 3.中国科学技术大学计算机科学与技术学院,安徽合肥,230027
  • 出版日期:2015-07-30 发布日期:2023-05-15
  • 通讯作者: 郑启龙,博士/副教授.
  • 作者简介:王向前,男,1985年生,博士生. 研究方向:编译优化与系统软件优化.

An in-place FFT algorithm for high data parallelism architecture

  1. 1.School of Computer and Information, Hefei University of Technoogy, Hefei 230009, China;  2. No.38 Research Institude, China Electronics Technoogy Group Corporation, Hefei 230088, China;  3. School of Computer Science and Technology, University of Science and Technology of China, Hefei 230027, China
  • Online:2015-07-30 Published:2023-05-15

摘要: 数字信号处理器的内存较小,而且数字信号处理领域的应用往往是数据密集型,这要求在设计数字信号处理应用算法时既要考虑时间复杂度又要兼顾算法的空间复杂度.为此提出了一种原位的逆序算法;针对数字信号处理器比较高的内存访问并行度,设计了部分逆序的原位高效FFT算法;并在魂芯DSP平台上实现了该算法框架.实验表明,与非原位FFT算法相比,该原位算法的空间复杂度大幅降低而时间效率的损失在可接受范围之内.

关键词: 逆序, 原位, FFT, 空间复杂度, 时间复杂度

Abstract: The on-chip memory in DSP is small, and applications for DSP are often data-intensive, which requires that space complexity as well as time complexity must be considered when algorithms are designed. So a in-place bit reverse algorithm was proposed. Then, to take advantage of memory bandwidth offered by DSP, an effective in-place FFT algorithm with part bit reverse was designed and implemented on BWDSP. Experiment result shows that, compared with the out-of-place FFT algorithm, its space complexity is significantly reduced,while the loss of time efficiency for the proposed in-place FFT algorithm is acceptable.

Key words: bit reverse, inplace, FFT, space complexity, time complexity

中图分类号: