home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.parallel
- Path: sparky!uunet!gatech!hubcap!fpst
- From: fineberg@nas.nasa.gov (Samuel A. Fineberg)
- Subject: Re: what is MPP-256(1Q93)
- Message-ID: <1993Jan21.135944.5045@hubcap.clemson.edu>
- Apparently-To: comp-parallel@ames.arc.nasa.gov
- Sender: news@nas.nasa.gov (News Administrator)
- Reply-To: fineberg@nas.nasa.gov
- Organization: CSC, NASA Ames Research Center, NAS Division
- References: <1993Jan20.124758.8100@hubcap.clemson.edu>
- Date: Wed, 20 Jan 93 23:37:03 GMT
- Approved: parallel@hubcap.clemson.edu
- Lines: 40
-
- In article <1993Jan20.124758.8100@hubcap.clemson.edu>, jtrsmith@garnet.berkeley.edu () writes:
- |> Hi, Sir,
- |> I am not familiar with supercomputers, although I have used Cray XMP
- |> in the campus.
- |>
- |> I have following questions about supercomputers, if you can give me
- |> any ideas, I should be grateful.
- |>
- |> (1) one is the subject, what is MPP-256(1Q93), who makes it?
- |> (2) we have a account on CM-5 in the campus
- |> We are now trying to change our Fortran programs to be used in the
- |> machine. The program is used to run on the campus Cray which is gone
- |> because of tight budget. The program is for large scale
- |> computational fluid dynamics.
- |> My second question is:
- |> Is it worth changing the program for the use on CM-5? Because I
- |> am not sure the direction of supercomputers. I am afraid that after
- |> the change, the program can not be used on other machines, such
- |> as Cray, and other parallel machines, like n-Cube. Then it will
- |> waste our time and money.
- |> (3) I know the FORTRAN of Cray is very close to Fortran used on small
- |> machines. But I dont know the FORTRANs on these parallel machines
- |> like, CM, n-cube, etc, are portable, are they?
- |>
- |> Thanks.
- |> John
- |>
- If you have the vector units, it is probably worth trying the CM5. It is
- relatively easy to do a painless (read slow) port from a Cray code to CM-FORTRAN
- that will perform better than your average workstation. How much otimization
- you want to do, is another story. If you stick to FORTRAN-90 or HPF extensions
- of FORTRAN, you won't have to do much to port it to future systems (virtually
- all new systems claim to be supporting HPF including Intel, Cray, Convex, KSR,
- etc.), though you probably won't get optimal performance.
-
- Sam
-
- P.S., an MPP-256 is Cray's parallel machine and I wouldn't worry too much about
- writing code that will work on an nCUBE, they probably won't be around for long.
-
-