Something very strange with ColChess

[Colin Frayn]

Geschrieben von: / Posted by: Colin Frayn at 09 May 2000 10:54:33:
There's something strange with the latest version of ColChess. Despite the fact that it performs well in all the tests I tried it on, it still seems to play very badly in short blitz games, especially on slower computers.
I have no idea why this is, but I'm currently looking into it. Hopefully I will have it resolved soon-ish.
Cheers,
Colin

ColChess Homepage

[Colin Frayn]

Geschrieben von: / Posted by: Colin Frayn at 10 May 2000 12:10:22:
Als Antwort auf: / As an answer to: Something very strange with ColChess geschrieben von: / posted by: Colin Frayn at 09 May 2000 10:54:33:
OK folks - I think I might have fixed the problem. Or at least I removed a few old bits of code and added in a few new bits and ColChess seems to now be getting reasonable results, performing about 30 ELO beneath Sjeng on 5 minute Blitz on my vocomputer.
New source and exe on my webpage.
Let me know if it's playing properly now.
Cheers,
Colin

ColChess Homepage

[Mogens Larsen]

Geschrieben von: / Posted by: Mogens Larsen at 10 May 2000 12:29:00:
Als Antwort auf: / As an answer to: Problem (hopefully) fixed geschrieben von: / posted by: Colin Frayn at 10 May 2000 12:10:22:

OK folks - I think I might have fixed the problem. Or at least I removed a few old bits of code and added in a few new bits and ColChess seems to now be getting reasonable results, performing about 30 ELO beneath Sjeng on 5 minute Blitz on my vocomputer.
New source and exe on my webpage.
Let me know if it's playing properly now.

I'll try repeating my tests with the new executable and see what happens.
Sincerely,
Mogens

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 10 May 2000 15:29:52:
Als Antwort auf: / As an answer to: Problem (hopefully) fixed geschrieben von: / posted by: Colin Frayn at 10 May 2000 12:10:22:

OK folks - I think I might have fixed the problem. Or at least I removed a few old bits of code and added in a few new bits and ColChess seems to now be getting reasonable results, performing about 30 ELO beneath Sjeng on 5 minute Blitz on my vocomputer.
New source and exe on my webpage.
Let me know if it's playing properly now.
Cheers,
Colin

Hi, Colin,
A quick match: Averno 0.28 - ColChess 6.2.1 7-3 (6-2-2). Pentium 166 with 32 MB RAM, WB 4.0.5, game in 5 minutes, 4 MB hash.
I have Winboard.debug which I can send you and José Carlos if you wish.
Gábor

[Mogens Larsen]

Geschrieben von: / Posted by: Mogens Larsen at 10 May 2000 16:25:35:
Als Antwort auf: / As an answer to: Re: Averno-ColChess geschrieben von: / posted by: Gabor Szots at 10 May 2000 15:29:52:

OK folks - I think I might have fixed the problem. Or at least I removed a few old bits of code and added in a few new bits and ColChess seems to now be getting reasonable results, performing about 30 ELO beneath Sjeng on 5 minute Blitz on my vocomputer.
New source and exe on my webpage.
Let me know if it's playing properly now.
Cheers,
Colin

Hi, Colin,
A quick match: Averno 0.28 - ColChess 6.2.1 7-3 (6-2-2). Pentium 166 with 32 MB RAM, WB 4.0.5, game in 5 minutes, 4 MB hash.
I have Winboard.debug which I can send you and José Carlos if you wish.

Quick match as well: ColChess 6.2.1 - Averno 0.28: 7½-12½ (7-12-1). PII-266MHz/64Mb, approx. 16Mb hash, WB 4.0.7, Game/5.
A slight improvement, but not enough for a conclusion of course. Games and debug available.
Sincerely,
Mogens

[Colin Frayn]

Geschrieben von: / Posted by: Colin Frayn at 10 May 2000 18:21:06:
Als Antwort auf: / As an answer to: Re: Averno-ColChess geschrieben von: / posted by: Mogens Larsen at 10 May 2000 16:25:35:

OK folks - I think I might have fixed the problem. Or at least I removed a few old bits of code and added in a few new bits and ColChess seems to now be getting reasonable results, performing about 30 ELO beneath Sjeng on 5 minute Blitz on my vocomputer.
New source and exe on my webpage.
Let me know if it's playing properly now.
Cheers,
Colin

Hi, Colin,
A quick match: Averno 0.28 - ColChess 6.2.1 7-3 (6-2-2). Pentium 166 with 32 MB RAM, WB 4.0.5, game in 5 minutes, 4 MB hash.
I have Winboard.debug which I can send you and José Carlos if you wish.

Quick match as well: ColChess 6.2.1 - Averno 0.28: 7½-12½ (7-12-1). PII-266MHz/64Mb, approx. 16Mb hash, WB 4.0.7, Game/5.
A slight improvement, but not enough for a conclusion of course. Games and debug available.

Yes please - if you could send me those games. (Both of you).
ColChess plays badly at very short time controls, or alternatively on slower computers. Not sure why this is, but I'm not too bothered as it soon plays better at longer controls. Middle game searching speed is still down unfortunately.
I always thought I'd do _slightly_ worse than Averno and Sjeng, so that's not a problem. As I keep saying, ColChess is not by any means optimised for playing games, it's optimised for solving problems accurately.
I'm running a 10 minute blitz game atm against Sjeng on my PIII-450 and the score is 12-4-15, so this indicates at most a 30-40 ELO point deficit.
Not a problem
I'll play a few longer game/40 games overnight and see what I get.
Cheers,
Colin

ColChess Homepage

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 11 May 2000 08:28:16:
Als Antwort auf: / As an answer to: Re: Averno-ColChess geschrieben von: / posted by: Colin Frayn at 10 May 2000 18:21:06:

[Colin Frayn]

Geschrieben von: / Posted by: Colin Frayn at 11 May 2000 19:26:12:
Als Antwort auf: / As an answer to: Something very strange with ColChess geschrieben von: / posted by: Colin Frayn at 09 May 2000 10:54:33:
I've just finished running a few matches against Sjeng and SSEChess to test ColChess' new version (6.2.1) and I got the following results; (Won-Drew-Lost)
ColChess v Sjeng (5/Blitz) 33-15-45 Estimated ELO Diff. = 45
ColChess v Sjeng (10/Blitz) 19-8-22 Estimated ELO Diff. = 21
ColChess v SSEChess (10/Blitz) 7-7-18 Estimated ELO Diff. = 124
I think that these correlate nicely (in fact might slightly improve upon) some of the existing results. I hope this makes sense.
These were on a P3-450Mhz. ColChess performs relatively better on faster computers and on longer time controls.
Cheers,
Colin

ColChess Homepage