Asri-unix.360 net.chess utzoo!decvax!ucbvax!menlo70!sri-unix!mclure@SRI-UNIX Tue Dec 29 11:49:21 1981 Limitations of brute-force programs Here is a reprint from an earlier submission to the mailing list which some of you may not have seen (Arpanet people in particular). From: sri-unix!mclure Newsgroups: net.chess Title: Beyond Belle Article-I.D.: sri-unix.268 Posted: Thu Dec 17 15:26:10 1981 Expires: Thu Dec 31 15:26:10 1981 Date: Thu Dec 17 15:26:10 1981 What is the limit of Belle-type programs? (e.g. full-width tree-searching algorithms implemented in hardware) Just how good can such programs get before the exponential explosion in the chess tree becomes too much for any hardware imaginable? I posed this question to Belle's author Thompson and his answer is interesting. Thompson feels that an "Ultra-Belle" is possible with the forthcoming generation of hardware involving Josephson junction switching and some of the other fast switching hardware which should be available in the next decade. He sees the following as the theoretical limit for an "Ultra-Belle": Ultra-Belle Belle ---------------- ---------------- 11-ply ave depth 8-ply ave depth in middle game 10^7 nodes/sec 10^5 nodes/sec 2350-2450 rating 2150-2250 rating So such a program would play at the IM or weak-GM level and be 100 times faster than Belle. The 3-ply increase in depth would only be enough for 200 points. ----------------------------------------------------------------- gopher://quux.org/ conversion by John Goerzen of http://communication.ucsd.edu/A-News/ This Usenet Oldnews Archive article may be copied and distributed freely, provided: 1. There is no money collected for the text(s) of the articles. 2. The following notice remains appended to each copy: The Usenet Oldnews Archive: Compilation Copyright (C) 1981, 1996 Bruce Jones, Henry Spencer, David Wiseman.