Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: a mistake of Fruit2.2Uri in CEGT (bug or another problem?)

Author: Heinz van Kempen

Date: 13:04:30 10/18/05

Go up one level in this thread


Hi Uri,

this was a very long game and I do not know what happens then with depth search
in the end. Maybe time trouble, maybe something else. I will try to reproduce.

I do not suspect a hardware error here, as especially in this match run by me
Fruit 2.2 got one of the best scores.

Additionally I can tell that there is no hash bug for this version. This was
checked many times. I have to tell that I am slowly getting tired to answer to
posts doubting in performance of the setting or the conditions, because it is
still not clear.

My goal is not to please you, although the history threshold reduce was your
idea, but to give the Fruit team clues what to adapt or not and it may well be
that history threshold reducement can be productive when testing only a few
positions. Maybe it will produce also bad moves when good moves were chosen with
history threshold=70. Did you also test this?

Overall I got more decent results for this setting this evening (still not
included in the list) and probably it will be close and comparable to default.

Best Regards
Heinz



This page took 0.01 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.