Computer Chess Club Archives


Search

Terms

Messages

Subject: New Winboard Engine

Author: Miguel A. Ballicora

Date: 07:13:44 01/20/01


My engine, "Gaviota" is available to be dowloaded at

http://www.msu.edu/~ballicor/gav

(this would be version 0.1)

It was written in ANSI C and compiled with djgpp. Later I found out that
it is supposed to give problems with winboard but I had no problems with
that compiler.

This is a starting point. The engine is still weak but much stronger
than I expected at this stage. I estimate is around Elo 2000 in an
AMD K6-2 400 mhz.
In FICS oscillates between 2000-2050. It had played around 40 games
there 5 0 or 3 3.
It is certainly weaker than Amyan (+1 -2 =1) by probably 100-150 points.
it is stronger than TSCP (in a match at 10 0 it was (+4 -1 =3).
In FICS looked a bit stronger than Sjeng version.
I was difficult to compare it to noonian. Lost several games the other day
but yesterday went (+4 -0 =0) in FICS.
Of course these are very few games bu just give an idea. I think that
~2000 could be a good estimation. Maybe 1900?

It plays weird. There are certain positions that handle reasonably well, but in
others is terrible. It seems that gaviota feels more comfortable in closed
positions.

It has no book, but it creates a 1mb file to learn bad positions (pos.bin)
In this way it does not repeat the same game over and over. It may play
bad openings, but improves.
It has 8 mb Hash table.
I believe that it is running stable. Let me know if there is a problem
using it with Winboard (I have been using 4.1). I does not support
"move now" but you can play normally with any time control and also matches
since supports "new". It does not support "edit".
It does not ponder.
Is this the first Argentinian Winboard Engine?

Regards,
Migue




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.