shankar@bedrock.SRC.Honeywell.COM (Son of Knuth) (04/05/89)
A functional program can be thought of as a tree, each node associated with an operation. A node is executed by first spawning all children nodes, waiting for each child to reply with an answer, and then executing the original parent node. For parallel execution, you simply spawn children in parallel. I am looking for studies on the effects of the tree shape (e.g. branching factor, branching factor distribution, node service time) on the resulting performance. --- Subash Shankar Honeywell Systems & Research Center voice: (612) 782 7558 US Snail: 3660 Technology Dr., Minneapolis, MN 55418 shankar@src.honeywell.com srcsip!shankar
shs@uts.amdahl.com (Steve Schoettler) (04/08/89)
In article <19971@srcsip.UUCP> shankar@bedrock.UUCP (Son of Knuth) writes: > >A functional program can be thought of as a tree, each node associated >with an operation. A node is executed by first spawning all children >nodes, waiting for each child to reply with an answer, and then executing >the original parent node. For parallel execution, you simply spawn >children in parallel. > >I am looking for studies on the effects of the tree shape (e.g. branching >factor, branching factor distribution, node service time) on the resulting >performance. >--- >Subash Shankar Honeywell Systems & Research Center >voice: (612) 782 7558 US Snail: 3660 Technology Dr., Minneapolis, MN 55418 >shankar@src.honeywell.com srcsip!shankar A very good analytic treatment of parallel geometries can be found in: "Multicomputer Networks: Message-Based Parallel Processing" by Daniel A. Reed and Richard M. Fujimoto, MIT Press, 1987. A good overview of the subject is in Chapter 6 (Multiprocessors) of "High Performance Computer Architecture" by Harold S. Stone, Addison Wesley, 1987. Steve -- Steve Schoettler shs@uts.amdahl.com {sun,decwrl,pyramid,ames,uunet}!amdahl!shs Amdahl Corp., M/S 213, 1250 E. Arques Ave, Sunnyvale, CA 94088