Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Glaurung 0.2.1

Author: Tord Romstad

Date: 17:43:32 02/09/05

Go up one level in this thread


Hi Norm and Brice,

I reply to your questions below, and I will also try to follow this thread a few
more days
and reply to any followup questions.  Generally it is better to ask questions in
the Winboard
Forum, which I read more often.  If you post here, there is a risk that I will
not see it.

On February 09, 2005 at 19:22:20, Norm Pollock wrote:

>On February 09, 2005 at 17:35:07, Brice BOISSEL wrote:
>
>>Are you continue to working on Gothmog ?

Yes, to a small extent.  At least I intend to release a final 1.0 version,
probably very soon.
After that, I am not sure.  I will definitely spend most of my time working on
Glaurung and
Scatha (Glaurung's twin brother, who plays hexagonal chess), but there is a
small possibility
that I will want to test whether some of the new ideas I get while working on
Glaurung work
in Gothmog as well.  If the results seem interesting, it is possible that there
will be new
Gothmog versions even after 1.0.

>>Is Glaurung the successor of Gothmog ?

Yes.  Gothmog was becoming too big, complicated and buggy, and it was very
frustrating
to work with the old and rotten code.  In the end, the only possibilities were
to start with
a new program or to find a better hobby.

>>Is Glaurung 0.21 stronger than Gothmog ?

Glaurung is almost certainly stronger at blitz and bullet time controls.  I
think the same is true
at slower time, but because I have very little data it is difficult to be
certain.

Neither engine will ever be very strong, though (although I have some hope that
Glaurung
will eventually become one of the strongest open source engines).  Raw playing
strength
has never been (and never will be) my main goal.  I am more interested in
producing an
interesting and entertaining style of play (and in this respect, Gothmog is
perhaps still the
better of my two programs).

In Glaurung, I have the additional goal of making everything as small, simple
and
straightforward as possible.  One of my reasons for making it open source is to
serve
as an example for beginners.  I want to show that writing a chess program does
not
have to be difficult. You can do everything in the most straightforward and
obvious
way you can imagine, without worrying about speed or advanced chess knowledge,
and still achieve a playing strength of well above 2000.

Those who want a strong engine should look elsewhere.  Glaurung is for those who
likes to have every available UCI engine, those who wants something unsually
aggressive, and those who would like to see that even slow, simple and stupid
programs can play decent chess.

>Another question.
>
>Do Glaurung and Gothmog share code? If so, how much?

They share very little code.  Less than 1%, I would guess, and nothing at all in
the
chess-related parts of the program.  The few small pieces of shared code are
found in
interface functions (like polling for user input during the search).

On the other hand, there are several important parts of Glaurung's code which
can be
considered as "translations" of the corresponding parts of Gothmog.  Using
exactly the
same code is usually not possible, because of different data structures and
design goals
in the two programs.  The most important examples of such translations are the
evaluation of king safety and passed pawns, and the static exchange evaluation.
In all
cases, the code in Glaurung has been designed to be simpler, shorter and use
less
memory than in Gothmog.

Tord




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.