On 24 Jun., 18:45, Robert Bruce Carleton <[EMAIL PROTECTED]> wrote: > I've been looking at distributed objects in GNUstep to do distributed > computing. I tried out the example programs in the DO tutorial (they > work) and I've done some background reading. > > I'm curious if anyone has done some kind of comparison of Step style > DO with HPC message passing, such as MPI and PVM. On the surface, > they sound similar but they seem to have different audiences. > > I did some searching with Google and friends but I haven't turned up > anything like that yet. > > Any pointers or comments would be helpful.
I have looked a little at http://rac.uits.iu.edu/hpc/mpi_tutorial/s1_roundrobin.html and indeed it has similarities. The reason is that *Step DO API is sort of a high-level object&message passing wrapper for basic Mach messaging. And basically you always have to do the same steps: 1. establish a connection 2. queue data to be sent to the other side 3. handle incoming data Everything is synchronized in some event queueing mechanism (NSRunLoop). -- hns _______________________________________________ Discuss-gnustep mailing list [email protected] http://lists.gnu.org/mailman/listinfo/discuss-gnustep
