Page 1 of 2

new/updated engines: 2008/11/24

PostPosted: 25 Nov 2008, 07:09
by Ron Murawski
new/updated engines: 2008/11/24


Crafty 22.5
This version fixes a memory leak problem
source: ftp://ftp.cis.uab.edu/pub/hyatt/source/
Jim Ablett build: http://homepages.tesco.net/henry.ablett/jims.html
no Peter Skinner build yet

Chronos* 1.9 by Guillermo Filia
New engine.
http://sites.google.com/site/chronosce/Home
*The original engine named 'Chronos' is by Pawel Koziol & Edmund Moshammer!
I wish programmers would check for name conflicts before naming their engines.
I am calling this engine 'Chronos*'


Additional information:
Computer-chess Wiki: Chess Engine List



Other computer-chess news


2008/11/23
Chess for Android phone
http://aartbik.blogspot.com/


Additional information:
Computer-chess Wiki: Chess News List

Post edited to remove a limitation listed against Android Chess. This limitation has been subsequently corrected.

Re: new/updated engines: 2008/11/24

PostPosted: 25 Nov 2008, 08:08
by Guenther Simon
Ron Murawski wrote:new/updated engines: 2008/11/24


Chronos* 1.9 by Guillermo Filia
New engine.
http://sites.google.com/site/chronosce/Home
*The original engine named 'Chronos' is by Pawel Koziol & Edmund Moshammer!
I wish programmers would check for name conflicts before naming their engines.
I am calling this engine 'Chronos*'




I wish he would also name the program started with, if it is true at all...

From the 'history.txt', just 6.5 months ago:

#
01/05/08:

*
Project starts from a previous non-bitboard and weak engine.


Guenther

Re: new/updated engines: 2008/11/24

PostPosted: 25 Nov 2008, 10:12
by Pedro Castro
Guenther Simon wrote:
Ron Murawski wrote:new/updated engines: 2008/11/24


Chronos* 1.9 by Guillermo Filia
New engine.
http://sites.google.com/site/chronosce/Home
*The original engine named 'Chronos' is by Pawel Koziol & Edmund Moshammer!
I wish programmers would check for name conflicts before naming their engines.
I am calling this engine 'Chronos*'




I wish he would also name the program started with, if it is true at all...

From the 'history.txt', just 6.5 months ago:

#
01/05/08:

*
Project starts from a previous non-bitboard and weak engine.


Guenther


Project starts from a previous non-bitboard and weak engine named GFC ("Guillermo Filia Chess"). Source code can be found in Downloads section.

Re: new/updated engines: 2008/11/24

PostPosted: 25 Nov 2008, 10:24
by guillef
I've updated the "history" and "downloads" sections. I have included some source code for my first two programs.

Best regards!

Re: new/updated engines: 2008/11/24

PostPosted: 26 Nov 2008, 05:00
by Dann Corbit
guillef wrote:I've updated the "history" and "downloads" sections. I have included some source code for my first two programs.

Best regards!


In your old program gfcc v2, you sometimes index arrays using a comma operator:

arr[a,b]

That is not correct. It needs to be
arr[a][b]

Otherwise the compiler interprets it simply as the second value and address as if a 1-d array. Probably not what you want.

Re: new/updated engines: 2008/11/24

PostPosted: 26 Nov 2008, 06:33
by guillef
Dann, thank you very much!.
During this period, building these engines, one of the challenges that cost me more to overcome was to know the language C. This is a task that is still giving me a lot of work because I was never dedicated to studying this language seriously, before having to use it for my projects.

With reference to the code, I tell you that I am currently trying to bypass multi dimensional arrays, as far as possible using one-dimensional arrays, accessing these through indices generated according to the case:

history_heuristics [64 * from + to] += 1 <<depth;

I believe this is better (and faster) than the multi-dimensional arrays, it is true?

Finally, thanks to fix those bugs in the engine gfc v2!!!
Guille.

Re: new/updated engines: 2008/11/24

PostPosted: 26 Nov 2008, 10:08
by Teemu Pudas
guillef wrote:history_heuristics [64 * from + to] += 1 <<depth;

I believe this is better (and faster) than the multi-dimensional arrays, it is true?


How could it possibly be faster? It's just different syntax for the same thing.

Re: new/updated engines: 2008/11/24

PostPosted: 26 Nov 2008, 12:40
by guillef
Some papers on computer chess specify in their examples, almost exclusively, implementations with one-dimensional arrays, as the history Heuristics papers, for example. Always felt that this was a best-practice to improve the overall performance of the algorithm.

Re: new/updated engines: 2008/11/24

PostPosted: 27 Nov 2008, 18:53
by Edmund
Ron Murawski wrote:[...]
Chronos* 1.9 by Guillermo Filia
New engine.
http://sites.google.com/site/chronosce/Home
*The original engine named 'Chronos' is by Pawel Koziol & Edmund Moshammer!
I wish programmers would check for name conflicts before naming their engines.
I am calling this engine 'Chronos*'
[...]


I have now agreed with Pawel to rename our Chronos to Glass. So that this confusion may end and we don't need two different engines in the rating list which only get distinguished by an asterisk.

Guillermo, its just unfortunate that you and I seem to have a similar imagination.

Re: new/updated engines: 2008/11/24

PostPosted: 27 Nov 2008, 19:45
by Don Cross
I use one-dimensional arrays in my chess program (Chenard) because it is far simpler to manage a single coordinate than x and y independently. I also use 12x12 board, with the 8x8 board inside a "belt" of 2-squares thickness for "off board" squares that are filled with a special value, so knights can't leave the board accidentally. For example, a knight moving two squares to the right, and one square up (e.g. from a1 to c2) is represented by the offset increasing by +14. So the code to see if a white knight can move in that direction would be something like:

Code: Select all
if ((board[offset + 14] & (OFFBOARD_BIT | WHITE_BIT)) == 0) {
    // white knight can move to that square
}

Re: new/updated engines: 2008/11/24

PostPosted: 27 Nov 2008, 20:05
by H.G.Muller
Edmund wrote:... we don't need two different engines in the rating list which only get distinguished by an asterisk.

Especially since there already is an engine called Astersisk... :D

Re: new/updated engines: 2008/11/24

PostPosted: 27 Nov 2008, 20:49
by guillef
Edmung, I want to thank you and Pavel for accepting change the name of your engine.I also had a second choice of name for my engine, this was "Havac", taken from the legendary racing game "Rock'n roll racing." The "Havac" was the most powerful car!. If anyone would like that name ... I think that is available for now.

Re: new/updated engines: 2008/11/24

PostPosted: 27 Nov 2008, 21:19
by Olivier Deville
Hi Don

By the way, do you have any version that I can test for next OpenWar ? Do the latest releases support incremental time control ?

Olivier

Re: new/updated engines: 2008/11/24

PostPosted: 27 Nov 2008, 21:59
by Don Cross
Hi Olivier,

It is good to hear from you again.

I just posted the latest version: 2008.11.27 (bug fix for extremely rare crash where Chenard tried to free a pointer that was not initialized). I have been testing the code with this fix over the past several days and the engine seems pretty stable now. I am going to try to avoid messing with it any more until the contest!

And yes, the WinBoard version of Chenard now understands incremental time controls and pondering. Here is the engine in a zip file:

http://cosinekitty.com/chenard/wxchenard.zip

By the way, I also have a fairly large (13.5 MB zipped) opening/experience library that I would like to include in the contest, but I am still tweaking it a bit. Will such a large library be a problem for this contest?

Thanks,

- Don

Re: new/updated engines: 2008/11/24

PostPosted: 27 Nov 2008, 22:10
by Olivier Deville
Olivier Deville wrote:Hi Don

By the way, do you have any version that I can test for next OpenWar ? Do the latest releases support incremental time control ?

Olivier


Thanks Don, just got the wb version :)

No problem with the large opening/learning file.

Olivier

Re: new/updated engines: 2008/11/24

PostPosted: 28 Nov 2008, 06:38
by Ron Murawski
Edmund wrote:
Ron Murawski wrote:[...]
Chronos* 1.9 by Guillermo Filia
New engine.
http://sites.google.com/site/chronosce/Home
*The original engine named 'Chronos' is by Pawel Koziol & Edmund Moshammer!
I wish programmers would check for name conflicts before naming their engines.
I am calling this engine 'Chronos*'
[...]


I have now agreed with Pawel to rename our Chronos to Glass. So that this confusion may end and we don't need two different engines in the rating list which only get distinguished by an asterisk.



I have updated the Private Engine List.
Computer-chess Wiki: Private Engine List
I am very happy the naming conflict has been resolved! I will remove the asterisk.

To all chess programmers: The names on the Private Engine List are *reserved* names. Please do not use them!

Ron

Re: new/updated engines: 2008/11/24

PostPosted: 30 Nov 2008, 18:22
by Pablo
Ron Murawski wrote:new/updated engines: 2008/11/24

Hello Ron,
At the site of Guillermo Filia, Chronos author, appear also the source codes of another machine called GFC, Itried to compile it but apparently is missing a file called "data.h".
I wonder if you have the mail adress of G.Filia, so to try to clear with him the issue of GFC source codes.
Best regards,
Pablo Urzua


Crafty 22.5
This version fixes a memory leak problem
source: ftp://ftp.cis.uab.edu/pub/hyatt/source/
Jim Ablett build: http://homepages.tesco.net/henry.ablett/jims.html
no Peter Skinner build yet

Chronos* 1.9 by Guillermo Filia
New engine.
http://sites.google.com/site/chronosce/Home
*The original engine named 'Chronos' is by Pawel Koziol & Edmund Moshammer!
I wish programmers would check for name conflicts before naming their engines.
I am calling this engine 'Chronos*'


Additional information:
Computer-chess Wiki: Chess Engine List



Other computer-chess news


2008/11/23
Chess for Android phone
http://aartbik.blogspot.com/


Additional information:
Computer-chess Wiki: Chess News List

Post edited to remove a limitation listed against Android Chess. This limitation has been subsequently corrected.

Re: new/updated engines: 2008/11/24

PostPosted: 02 Dec 2008, 14:50
by guillef
Pablo, i'll search this missing file and i'll upload the corrected sources as soon as posible. By the way, the sources in GFC v2.zip are more recent and compile well!, and the engine is a little better in that version (about 200+ elo).

Guille.

Re: new/updated engines: 2008/11/24

PostPosted: 02 Dec 2008, 21:05
by Roger Brown
Don Cross wrote:
By the way, I also have a fairly large (13.5 MB zipped) opening/experience library that I would like to include in the contest, but I am still tweaking it a bit. Will such a large library be a problem for this contest?

Thanks,

- Don




Hello Don Cross,

For the less important folk such as non-tournament persons like me, can this large book be made available somewhere? I like the engines to play with their own books as far as possible.

Pretty please?

Later.

Re: new/updated engines: 2008/11/24

PostPosted: 04 Dec 2008, 19:58
by guillef
Hola Pablo, traté de escribirte a la direccion que me pasaste desde gmail pero el mail me retorno con error de envío. Podes escribirme directamente a guillef@gmail.com

Saludos.