Newsgroups: comp.parallel.mpi From: engrbohn@aol.com (Engr Bohn) Subject: Re: NOW vs MPI Organization: AOL http://www.aol.com Date: 06 May 1998 10:59:14 GMT Message-ID: <1998050610591400.GAA24259@ladder03.news.aol.com> Tom Impelluso wrote: >I have written a code using MPI distributed over a cluster of >workstations. > >Someone told me to look at NOW (Network of Workstations). > >OK, I did but I cannot see why I should prefer NOW over MPI? > >Could someone enlighten me as to what NOW is and how it differs from >MPI? Hmm. I've been using the terms COW & NOW more-or-less interchangably. For example, when I use the machines that were networked together with a myrinet specifically for distributed/parallel research, I may call it a COW or a NOW, depending on whichever word comes to my mouth first. When I use the machines that were intended for general student use and are networked together with a thin ethernet, I might refer to it as a NOW, though I'd more likely call it "the Zoo" (all the machines are named after animals), but I'd have a hard time convincing myself it was a "cluster". Regardless of which system I use, I use MPI. Perhaps the someone who told you to look at NOW was referring to the Berkeley NOW project, which I've been told is doing research into such areas as cycle harvesting (implementation & policy) and global address space (using distributed memory directly and/or as swap space). I can certainly see how someone who was thinking of the Berkeley NOW project, with its shared global address space, as being different from message passing. cb Christopher A. Bohn EngrBohn@aol.com "Oooh! What does THIS button do!?"