Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16] [ 17] [ 18] [ 19] [ 20]

@

Journal of Information Science and Engineering, Vol. 23 No. 1, pp. 315-327 (January 2007)

Efficient Data Distribution Scheme for Multi-Dimensional Sparse Arrays

Chun-Yuan Lin and Yeh-Ching Chung*
Institute of Molecular and Cellular Biology
*Department of Computer Science
National Tsing Hua University
Hsinchu, 300 Taiwan
E-mail: ychung@cs.nthu.edu.tw

Array operations are useful in a large number of important scientific codes, such as molecular dynamics, finite-element methods, climate modeling, etc. It is a challenging problem to provide an efficient data distribution for irregular problems. Multi-dimensional (MD) sparse array operations can be used in atmosphere and ocean sciences, image processing, etc., and have been an extensively investigated problem. In our previous work, a data distribution scheme, Encoding-Decoding (ED), was proposed for twodimensional (2D) sparse arrays. In this paper, ED is extended to be useful for MD sparse arrays first. Then, the performance of ED is compared with that of Send Followed Compress (SFC) and Compress Followed Send (CFS). Both theoretical analysis and experimental tests were conducted and then shown that ED is superior to SFC and CFS for all of evaluated criteria.

Keywords: data compression method, data distribution scheme, data layout, distributed memory multicomputers, sparse ratio, sparse array

Full Text () Retrieve PDF document (200701_19.pdf)

Received January 6, 2005; revised May 25 & September 28, 2005; accepted November 2, 2005.
Communicated by Gen-Huey Chen.