Computer Chess Club Archives




Subject: Re: is the

Author: Tom Kerrigan

Date: 09:59:38 08/05/98

Go up one level in this thread

Whoa, there, Tex...

Don't you remember why we started this conversation? It's because you told every
Tom, Dick, and Jane that 128 MB RAM is a necessity.

Run my experiment to prove that it is (or isn't!) necessary.

THEN worry about other experiments.


On August 05, 1998 at 11:34:39, Komputer Korner wrote:

>This is only going to tell us the speedup available up to 128 Mb of RAM which
>isn't enough. The hash table will get filled up too quickly and even with a good
>hash replacement strategy, there still will be a significant slowdown in the
>search. For overnight analysis we need 2Gb of RAM and a slow searcher like
>Hiarcs or CSTal. Fritz is too fast for this experiment. What we need to do is
>run the positions with hash tables until the hash table fills  with successive
>increasing amounts of hash tables until you reach 2Gb of hash tables. Each
>experiment is to be run until the hash table is filled on each position and then
>rerun the same position without hash tables. Check the number of nodes/ply depth
>reached. Each program and machine combination will be different of course. Make
>sure that  the 2 scenarios are on a representative mix of endgame and middlegame
>positions. Are there any slow programs that can accomodate 2Gb of hash tables.
>Are there any machines that can accomodate 2Gb of RAM?
>Komputer Korner

This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.