An adaptable cluster structure of (sm)2 -II
In analyzing electronic circuits, it is usually necessary to solve a sparse coefficient matrix comprised of simultaneous linear equations. In order to treat these problems effectively, we have developed a dedicated parallel machine called (sm)2-II (’the Sparse Matrix Solving Machine’ version II).
(sm)2-II is composed of multiple clusters, each consisting of multiple PUs (Processing Units) connected by a special communication mechanism called RSM (Receiver Selectable Multicast).
Here, we propose an adaptive cluster structure and evaluate its performance. The characteristics of typical problems are analyzed at the beginning of computation. Then the size of each cluster and the connections between clusters are adapted to the problem automatically. Using this mechanism, effective computation is possible in various types of problems.
Unable to display preview. Download preview PDF.
- R.Kober, "The Multiprocessor System SMS201 Combining 128 Microprocessors to a Powerful Computer," pp.225–229 Digest of Papers Compcon Fall 1977.Google Scholar
- R.J. Swan, et al., "The implementation of the Cm* multi-microprocessor," Proc. of NCC, Vol 46, pp.645–655, 1977.Google Scholar
- D. Gajski, et al., "Cedar-a large scale multiprocessor," Proc. of the International Conference on Parallel Processing, pp.524–529 Aug. 1983.Google Scholar
- T.Kudoh, et.al., "NDL:A language for solving scientific problems on MIMD machines," Proc. of 1st Int. Conf. on Super Computing Systems Dec.1985 pp.55–64Google Scholar
- H.Amano, et.al., "(sm)2-II: The new version of the Sparse Matrix Solving Machine," Proc. of 12TH annual International Symposium on Computer Architecture, June 1985Google Scholar