astfalk@uunet.UU.NET (Greg Astfalk) (09/25/89)
I'd like to get a precise definition, or reference to the definition, of what constitutes a 'Class 7' machine. I assume it is given in terms of MFLOPS and memory. Any pointers appreciated. Greg Astfalk 301-345-2400 Convex Computer Corp convex!astfalk 7501 Greenway Center Dr astfalk@convex.com Greenbelt, MD 20770-3514 na.astfalk@na-net.stanford.edu
eugene@eos.arc.nasa.gov (Eugene Miya) (09/26/89)
In short: This is none. Any person who will quote you sizes or speeds: Robins Williams recently said: If you can remember Woodstock, you weren't there. Numerical classification is an artifact of what is now the Dept. of Energy. The person you want to speak to who developed it is Sid Fernbach, formerly at LLL now a consultant to CDC. You had best speak soon as Sid is having health problems. He lives in San Jose. Classes are a numeric categorization of current fastest supercomputer. Hence, if a machine isn't a supercomputer (sic) then it doesn't rate. E.g., a VAX is many times more "powerful" [what ever that measurement is] than say many early computers, but there is no temporal basis for comparison. Hence the fastest class 6 machines were Crays, CDCs (earlier), NEC, Fujitsu, Hitachi. None of the "mini-supers" rated on this scale. "Classes" stopped growing about 1984 when the Cray-2 was released. I have pressed and pressed Sid for a definition of class 7 and whether examples for such exist. It's not clear. Further, machines like the connection machine and hypercube complicate the issue. Any person who can give simple definitions to what a Class 7 machine is only fooling himself and does not understand the issues....... yet. --eugene
brooks@vette.llnl.gov (Eugene Brooks) (09/26/89)
In article <6573@hubcap.clemson.edu> convex!convex1!astfalk@uunet.UU.NET (Greg Astfalk) writes: > >I'd like to get a precise definition, or reference to the >definition, of what constitutes a 'Class 7' machine. I >assume it is given in terms of MFLOPS and memory. Any pointers >appreciated. >From Eugene Miya's multiprocessing bibliography, it might have further pointers %A G. R. Allen %T Parallel Processing System -Cyberplus- %B Supercomputers, Class VI Systems, Hardware and Software %E Sidney Fernbach %I North-Holland %D 1986 %P 169-181 brooks@maddog.llnl.gov, brooks@maddog.uucp