Agreed. But it is one tool in J which explicitly shows how data can be partitioned. It is not the only tool which partitions. / /. \ \. @ and ;. also provide information as to how data can be split over multiple processors. The problem right now is that inter processor communication is slow and it is only practical to spread the work around on very large data sets, particularly if the processors do not share memory.
And issues such as processors shared over the Internet go far beyond this type of sharing. But now dual and quad processors are on laptops. If communication were as fast as say an add or multiply data these tools could spread J over multiple processors quite nicely. On Tue, Apr 28, 2009 at 10:18 PM, Steven H. Rogers <st...@shrogers.com> wrote: > Rank can be used effectively for programming simple cellular automata, > but actor models are usually more complex than that. It would certainly > be possible to take this approach, but that doesn't make it a good fit > for general actor models. > > # Steve > > Don Guinn wrote: >> It explicitly specifies independence between parts of the data. >> >> On Tue, Apr 28, 2009 at 8:07 PM, Steven H. Rogers <st...@shrogers.com> wrote: >> >>> Don Guinn wrote: >>> >>>> J has an explicit concurrency operator, it's called rank. >>>> >>> True, but how is that relevant to the Actor model? >>> >>> ---------------------------------------------------------------------- >>> For information about J forums see http://www.jsoftware.com/forums.htm >>> >>> >> ---------------------------------------------------------------------- >> For information about J forums see http://www.jsoftware.com/forums.htm >> >> > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
