Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: To Fabien Letouzey - about Toga

Author: Stephen A. Boak

Date: 21:53:46 03/17/05

Go up one level in this thread


On March 17, 2005 at 04:34:51, Gabor Szots wrote:

>On March 17, 2005 at 03:52:24, Stephen A. Boak wrote:
>
>>
>>I downloaded & ran the Toga UCI engine to analyze a tournament game of mine.
>>Previously I had run Shredder 9 UCI at 5 min / move.  This time I ran Toga at 5
>>min / move.
>>
>>Wow!!
>>
>>I was extremely impressed.
>>
>>For most moves, Toga matched Shredder 9 UCI suggested moves.
>>In perhaps one out of five or six moves, Toga suggested a different, but
>>excellent move.
>>
>>All in all, Toga is a marked improvement over Fruit alone, and is very powerful
>>and interesting for overnight analysis.
>>
>>Thanks to all authors involved in the original and modified engine codes.
>>
>>--Steve
>
>Hi Steve,
>
>Have you tried Fruit 2.0 for the same analysis? What was the result?
>
>Gábor

In this particular game I did not analyze with Fruit 2.0 (but I now will, and
will re-post with my findings).  But I have used it to analyze several other
games in the past.  It is certainly strong, and seems to analyze quickly to nice
depths.

My impressions of Fruit 2.0 (based on a limited number of comparisons of its
output with that of other programs I own) is that it is quite strong.

From what I read, it searches deeply (& quickly) but does not have much in the
way of eval in it.  Doesn't mean it is a weak engine, but only that it is less
knowledgeable about positional distinctions than other engines.

My feeling is that Fruit 2.0 is definitely not as positionally accurate as
Shredder 7.04 or Shredder 9 UCI (my favorite engines).

Toga appears to have added some significant knowledge, positional or analysis
improvements (my opinion, based on the above mentioned single game analysis).

As a test, I may create a table, comparing suggested moves & eval scores by
different engines for all moves in that same game.

The test trials would likely each be based on identical 512MB HT size and 5 min
/ move time provided to each engine.

After my test, we will see what move recommendations are different between Fruit
2.0 and Toga (as well as Shredder & Fritz engines), and thus we can judge for
ourselves the relative strengths of those two versions.

More later,
--Steve





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.