segall@vangogh.rutgers.edu (11/28/89)
I did not see the original article by Eugene Brooks, so I'm responding to this extract that appeared in Wen-King Su's reply. > In article <7130@hubcap.clemson.edu> brooks@maddog.llnl.gov writes: > < It is interesting to note that the orginal message > >passing machine at Caltech was proposed as a grid, not a hypercube. The > <Computer scientists there thought that hypercubes were prettier at the time > >and forced the physicists to learn Grey codes to get their physics problem > <mappings done. In Charles Lang's thesis (Caltech CS dept, early '80s, advisor Chuck Seitz), a number of interconnection topologies were investigated. These included simulations of a variety of topologies, including grids and hypercubes. Under the communication model they were assuming, the hypercube outperformed the grid network for a variety of problems, for reasons mentioned in Wen-King Su's posting. The choice was based on facts, not asthetics. Later developments provided a new communication method, and grids are better choice with that method. --Ed -- uucp: {...}!rutgers!caip.rutgers.edu!segall arpa: segall@caip.rutgers.edu