2018年12月10日月曜日

A paper on parallel algorithms for old supercomputers

As you may know, ResearchGate is one of social network services for academic persons. I have not used it very much, but I registered as a member in the past. Recently, the following e-mail arrived from ResearchGate.



The above paper was accepted in IEEE Transactions in 1985 (about 33 years ago). A kind of parallel computing algorithms for the supercomputer (vector machine) at the time was proposed and analyzed. This modest e-mail tells me that this paper is cited from 31 other papers and that 51 subscriptions have been made by ResearchGate readers. 

It is not a big event, but I am happy myself. At that time, the supercomputer was a vector computer with several number of parallel pipelines controlled by chaining technology. After that, there are thousands of processor configurations that are said to be massively parallel, and now they are transitioning to millions of many core (multi-core) configurations. However, underlying foundation of most parallel algorithms would be common to any configuration. For that reason, the algorithms for the old vector machine may still be a reference now.

0 件のコメント:

コメントを投稿