DOI QR코드

DOI QR Code

High-Performance FFT Using Data Reorganization

데이터 재구성 기법을 이용한 고성능 FFT

  • Published : 2005.06.01

Abstract

The efficient utilization of cache memories is a key factor in achieving high performance for computing large signal transforms. Nonunit stride access in computation of large DFTs causes cache conflict misses, thereby resulting in poor cache performance. It leads to a severe degradation in overall performance. In this paper, we propose a dynamic data layout approach considering the memory hierarchy system. In our approach, data reorganization is performed between computation stages to reduce the number of cache misses. Also, we develop an efficient search algorithm to determine the optimal tree with the minimum execution time among possible factorization trees considering the size of DFTs and the data access stride. Our approach is applied to compute the fast Fourier Transform (FFT). Experiments were performed on Pentium 4, $Athlon^{TM}$ 64, Alpha 21264, UtraSPARC III. Experiment results show that our FFT achieve performance improvement of up to 3.37 times better than the previous FFT packages.

대규모 신호처리 변환을 신속하게 처리하기 위해서는 캐시 메모리를 효과적으로 이용하는 것이 중요하다. 대규모 DFT 계산에서는 stride 액세스로 인한 캐시 충돌 적중 실패로 인하여 캐시 성능이 상당히 떨어지게 되고 이로 인해 전체적인 성능이 저하하게 된다. 본 논문에서는 메모리 계층 구조를 고려한 동적 데이터 재배열(Dynamic Data Layout) 방법을 개발하였다. 제시된 방법은 stride를 가지는 계산 단계(computation stage) 사이에 데이터를 동적으로 재구성을 하여 캐시 적중 실패를 줄이는 것이다. 또한 트리 구조 FFT 계산 방법에서 FFT 크기와 데이터 stride 액세스를 기초로 하여 가능한 모든 인수분해 트리 중에서 최소 실행시간을 가지는 최적의 인수 분해트리를 찾아내는 탐색 알고리즘을 개발하였다. 성능 향상을 확인하기 위하여 제시된 방법을 기존의 FFT 알고리즘에 적용하여 Pentium 4, Alpha 21264, $Athlon^{TM}$ 64, UltraSPARC III에서 실험하였다. 실험 결과에 따르면 기존의 FFT 패키지들과 비교하여 제시된 방법을 적용한 FFT가 최대 3.37배의 성능 향상을 얻을 수 있었다.

Keywords

References

  1. A. Ailamaki, D. DeWitt, M. D. Hill, M. Skounakis, 'Weaving Relations for Cache Performance,' in Proc. 27th International Conference Very Large Data Base, 2001
  2. D. H. Bailey, 'Unfavorable Strides in Cache Memory Systems,' Scientific Programming, 1995
  3. S. Egner, 'Zur Algorithmischen Zerlegungstheorie Linearer Transformationen mit Symmetire,' Ph.D. Thesis, Universitat Karlsruhe, 1997
  4. M. Frigo and S. G. Johnson, 'FFTW: An Adaptive Software Architecture for the FFT,' International Conference on Acoustics, Speech, and Signal Processing 1998 (ICASSP 1998), 3, 1998 https://doi.org/10.1109/ICASSP.1998.681704
  5. K. S. Gatlin and L. Carter, 'Faster FFTs via Architecture Cognizance,' International Conference on Parallel Architectures and Compilation Techniques (PACT2000), Oct., 2000 https://doi.org/10.1109/PACT.2000.888349
  6. G. Haentjens, 'An Investigation of Recursive FFT Implementations,' Master's Thesis, Dept. of Electrical and Computer Engineering, Canegie Mellon University, 2000
  7. J. Johnson and M. Piischel, 'In Search of the Optimal Walsh-Hadamard Transform,' International Conference on Acoustics, Speech, and Signal Processing 2000 (JCASSP 2000), June, 2000 https://doi.org/10.1109/ICASSP.2000.860117
  8. M. Linderman and R. Linderman, 'Real-Time STAP Demonstration on an Embedded High-Performance Computer,' National Radar Conference, 1997 https://doi.org/10.1109/NRC.1997.588132
  9. W. Liu and V. K. Prasanna, 'Utilizing the Power of High-Performance Computing,' IEEE Signal Processing, September, 1998 https://doi.org/10.1109/79.708542
  10. V. VanLoan, 'Computational Frameworks for the Fast Fourier Transform,' Frontieres in Applied Mathmetics, Vol. 10, SIAM, 1992
  11. D. Mirkovic, R. Mahassom, and L. Johnsson, 'An Adaptive Software Library for Fast Fourier Transforms,' Proceedings of the 2000 International Conference on Supercomputing, May, 2000
  12. N. Prak, B. Hong, and V. K. Prasanna, 'Analysis of Memory Hierarchy Performance of Block Data Layout,' Proceedings of the 2002 International Conference on Parallel Processing (JCPP 2002), Aug., 2002 https://doi.org/10.1109/ICPP.2002.1040857
  13. N. Prak and V. K. Prasanna, 'Cache Conscious Walsh-Hadamard Transform,' International Conference on Acoustics, Speech, and Signal Processing 2001 (ICASSP 2001), May, 2001 https://doi.org/10.1109/ICASSP.2001.941140
  14. R. Tolimieri, M. An, and C. Lu, 'Algorithms for Discrete Fourier Transforms and Convolution,' Springer, 1997