Asri-unix.370
net.chess
utzoo!decvax!ucbvax!menlo70!sri-unix!jim@RAND-UNIX
Wed Dec 30 13:25:23 1981
Re: Limitations of brute-force programs
I did some analysis for my dissertation at C-MU a few years ago, and
estimated that 13 ply was about the theoretical limit for my TECH program.
That involved about 400,000 processors in parallel, organized in a tree
structure with appropriate data passing; if memory serves, more would
have run into speed-of-light restrictions. That's probably similar to
Ken Thompson's 11 ply, since he does more processing at terminal nodes
than I did. I don't have a good estimate of the USCF rating for that
level, although I believe that it would find incredible tactical ideas
that are deep enough to look like strategies. In TECH's class D rating
range, I was still getting a linear increase in USCF rating with increases
in time, so that adding an extra 1/2 hour to the time control (from
50 moves in 120 minutes to 50 in 150, or 50 in 180) was worth another
80 USCF rating points. Presumably this would start topping out a few
rating classes farther up, so that the estimate would no longer be valid.
By the way, in TECH it took a factor of 25 increase in speed to search
an additional two ply.
Where are Ken's rating equations recorded?
Jim Gillogly
-----------------------------------------------------------------
gopher://quux.org/ conversion by John Goerzen <
[email protected]>
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.
