Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!unmvax!gatech!hubcap!halldors From: halldors@paul.rutgers.edu (Magnus M Halldorsson) Newsgroups: comp.parallel Subject: Re: Superlinear Message-ID: <3809@hubcap.UUCP> Date: 9 Dec 88 17:41:54 GMT Sender: fpst@hubcap.UUCP Lines: 25 Approved: parallel@hubcap.clemson.edu In article <3801@hubcap.UUCP> mcvax!ivax.doc.imperial.ac.uk!mmh@uunet.UU.NET (Matthew Huntbach) writes: > Depends on the algorithm. In parallel heuristic search superlinear > speedup can happen easily. >... > As a simple demonstration, suppose you are searching a tree with one branch > which your heuristic tells you is promising, but which after a lot of > computation (say 11 time units) does not yield a result, and another branch > which your heuristic tells you is less promising but which does in fact yield > a result after a small amount of computation (say 1 time unit). > > On a single processor you could search the whole large branch first > before turning to the small branch and finding your solution: total time > 12 units. But you can always simulate the parallel processors by traversing the tree breadth-first. That will require a stack and extra memory, but no more than the memory used by all the parallel processors combined. For this algorithm, the parallel version would be an example of AND-parallelism, while for the standard sequential algorithm, it would be an example of OR-parallelism. The moral is, you can only get superlinear speedup if the corresponding sequential algorithm you're comparing with is less optimal. Magnus