Performance evaluation of data distributions with load-balancing for sparse arrays

In our previous work, we have studied three data distribution schemes, Send Followed Compress (SFC), Compress Followed Send (CFS), and Encoding- Decoding (ED), for sparse arrays based on the traditional matrix representation (TMR) scheme. Since multidimensional arrays can also be represented by the extended Karnaugh map representation (EKMR) scheme, in this paper, we first apply the SFC/CFS/ED schemes based on the EKMR scheme. Then, we compare the performance of these three schemes with those based on the TMR scheme. Both theoretical analysis and experimental test were conducted. In theoretical analysis, we analyze the SFC/CFS/ED schemes based on the TMR/EKMR schemes in terms of the data distribution time and the data compression time. In experimental test, we implement these three schemes on an IBM SP2 parallel machine. The theoretical analysis and experimental results first show that the ED scheme is superior to the CFS scheme that is superior to the SFC scheme. Second, these three schemes based on the EKMR scheme outperform those based on the TMR scheme.