1. 西安邮电大学 机构 电子工程学院,陕西,西安,710121
2. 西北工业大学 计算机学院,陕西,西安,710072
纸质出版:2018
移动端阅览
常立博, 王时雨. 基于轻核阵列机的FFT算法并行化研究与实现[J]. 微电子学与计算机, 2018, 35(1): 100-105.
CHANG Li-bo, WANG Shi-yu. Research and Implementation of Parallel FFT Algorithm Based on Light Core Array[J]. 2018, 35(1): 100-105.
目前普遍采用基于流水的单路径延时反馈结构
或基于存储结构实现快速傅里叶变换(Fast Fourier Transform
FFT).前一种结构效率高但缺乏灵活性
而后一种结构通用性较好但性能较差.首先提出了一种FFT并行实现的算法
然后在并行图形阵列机(Parallel Array Architecture for Graphics
PAAG) 平台上实现了基2时间抽取的FFT (Decimation-In-Timer FFT
DIT-FFT) 算法; 最后将长度为512的DIT-FFT算法分别映射到1个PE、4个PE、8个PE和16个PE上实现.分析实验结果显示: 随着PE个数的增加
加速比呈曲线上升的趋势; 但随着PE个数的不断增加
算法执行速度增长开始变得缓慢
当映射到8个PE上时
最大加速比可以达到5.98.
At present
the fast Fourier transform is realized based on the flow-based single-path delay feedback or storage structure
the former structure is highly efficient but lacks flexibility
and the latter is more versatile but less robust.In this paper
firstly
an algorithm is implemented in parallel with FFT
and then the Decitation-InTimer FFT (DIT-FFT) algorithm is implemented on the Parallel Array Architecture for Graphics (PAAG) platform.Finally
the DIT-FFT algorithm with length of 512 is mapped to 1 PE
4 PE
8 PEs and 16 PEs respectively.The results show that with the increase of the number of PEs
the acceleration rate increases with the increase of the number of PEs.However
as the number of PEs increases
the speed of the implementation of the algorithm becomes slow
and when it is mapped to 8 PEs
The acceleration ratio can reach 5.98.
李涛, 杨婷, 易学渊, 等. 萤火虫2: 一种多态并行机的硬件体系结构[J]. 计算机工程与科学, 2014, 36(2): 191-200. DOI: 191
LI T. PAAG: A polymorphic array architecture for graphics and image processing[C]//PAAP'2012, IEEE Computer Society CPS. IEEE, 2012: 242-249.
Baas B M. A low-power, high-performance, 1024-Point FFT Processor[J]. IEEE Journal of Solid-State Circuits, 1999, 34(3): 380-387. DOI: 380
Huang S J, Chen S G. A high-throughput radix-16FFT processor with parallel and normal input/output ordering for IEEE 802.15. 3csystems[J]. Circuits % Systems Ⅰ Regular Papers IEEE Transactions on, 2012, 59(8): 1752-1765. DOI: 1752
李涛, 肖灵芝. 面向图形和图像处理的轻核阵列机结构[J]. 西安邮电大学学报, 2012, 17(3): 41-47. DOI: 41
Shin M, Lee H. A high-speed four-parallel radix-2 4, FFT/IFFT processor for UWB applications[C]//IEEE International Symposium on Circuits and Systems. IEEE, 2008: 960-963.
Chang Y N, Parhi K K. An efficient pipelined FFT architecture[J]. IEEE Transactions on Circuits % Systems Ⅱ Analog % Digital Signal Processing, 2003, 50(6): 322-325. DOI: 322
Cooley J W, Tukey J W. An algorithm for the machine calculation of complex Fourier series[J]. Mathematics of Computation, 1965, 19(19): 297-301. DOI: 297
丁玉美, 高西全. 数字信号处理[M]. 2版. 西安: 西安电子科技大学出版社, 2000.
方维, 孙广中, 吴超, 等. 一种三维快速傅里叶变换并行算法[J]. 计算机研究与发展, 2011, 48(3): 440-446. DOI: 440
Marowka A, Ramat Gan. Back to thin-core massively parallel processors[J]. IEEE Computer, 2011, 44(12): 49-54. DOI: 49
兰海洋, 林晓焕. 快速傅里叶变换的硬件实现[J]. 通信技术, 2012, 45(8): 115-117. DOI: 115
0
浏览量
8
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010602201714号