mfi@beach.cis.ufl.edu (Mark Interrante) (12/16/89)
I have come across a reference to a Paralation Model for parallel programming. I believe that it runs on top of lisp. Does anyone know about it? Is there a PD version for CommonLisp (MACL in particular)? Thanks, ----------------------------------------------------------------------------- Mark Interrante Software Engineering Research Center mfi@beach.cis.ufl.edu CIS Department, University of Florida 32611 ----------------------------------------------------------------------------- "Imagine what it would be like if TV actually were good. It would be the end of everything we know." Marvin Minsky
gary@think.com (Gary Sabot) (12/18/89)
I have come across a reference to a Paralation Model for parallel programming. I believe that it runs on top of lisp. Does anyone know about it? Is there a PD version for CommonLisp (MACL in particular)? There is no PD version, but the book described below contains a Tiny Paralation Lisp that you can type in, and the complete Paralation Lisp software on disk is only $100. Details: ------------ The paralation model is a model for parallel programming. It consists of a new data structure and a small number of operators, and can be combined with conventional base languages like Lisp and C. It is high-level (similar in some ways to SETL or APL), but it provides a compiler with lots of information for taking advantage of various features of the target architecture (MIMD, SIMD, hypercube, butterfly, etc.). Several compilers have been written for it; one compiled a recursive quicksort onto the Connection Machine so that it ran within a factor of 3 of the fastest (microcoded) sort on the CM. MIT Press is selling both a book about it, "The Paralation Model: Architecture-Independent Parallel Programming", and a simulator which includes full Lisp source code. The simulator runs under just about any Common Lisp you can find, including Lucid and Kyoto (under UNIX), and also Allegro, Symbolics, Gold Hill 3.1, etc. It is actually pretty fast. Some people have rewritten serial code into Paralation Lisp and found that both code size and runtime (on the same serial machine) go down. The reason is that the compiler is very smart about when it uses hash tables, array lookup tables, a-lists, etc, to represent aggregate data structures. MIT Press is (800) 356-0343 or (617) 253-2884. gary@think.com