MID-LOW Tournament

[Carlos E.A. Drake]

Geschrieben von:/Posted by: Carlos E.A. Drake at 02 March 2000 22:40:43:
Pentium 133 ~ 8 mhz HT
5 minutes + 2 seconds
Olithink 2.2.1 6 1/2
Crux 1.7 beta 5 1/2
Amyan 1.0 beta 5
TSCP 1.5 3 1/2
Averno 0.27 3 1/2
Snail 1.03 2
Colchess 5.6 1 1/2
Freyr 0.960 1/2
Olithink is unknown in test! Why?
Crux did best than Amyan in it test, but lower in other previous
TSCP with the command -st 5 play for best
Freyr is a pity...

[Colin Frayn]

Geschrieben von:/Posted by: Colin Frayn at 03 March 2000 12:11:42:
Als Antwort auf:/As an answer to: MID-LOW Tournament geschrieben von:/posted by: Carlos E.A. Drake at 02 March 2000 22:40:43:

Olithink 2.2.1 6 1/2
Crux 1.7 beta 5 1/2
Amyan 1.0 beta 5
TSCP 1.5 3 1/2
Averno 0.27 3 1/2
Snail 1.03 2
Colchess 5.6 1 1/2
Freyr 0.960 1/2

That's the strangest tournament results I've seen for a long time.
I'm surprised Amyan and Olithink did so well. I don't know about Crux. Freyr is a strong program so why that came bottom I have no idea. Averno should have come higher. No comment about ColChess. TSCP never fails to amaze me. I've taken the code apart time and time again, and there's simply no way it should play as well as it does, within an order of magnitude. How it came so high, I have no idea. ColChess' record against it is usually 90% wins, though TSCP rarely uses a third of ColChess' thinking time.
Did you use ColChess' large opening book? I think that is going to be the default book for version 6.0 as lots of people just use the really small example book and don't realise that there's a 400k one with it...
Cheers,
Col

ColChess Homepage

[WYx]

Geschrieben von:/Posted by: WYx at 03 March 2000 12:28:23:
Als Antwort auf:/As an answer to: MID-LOW Tournament geschrieben von:/posted by: Carlos E.A. Drake at 02 March 2000 22:40:43:

Pentium 133 ~ 8 mhz HT
5 minutes + 2 seconds
Olithink 2.2.1 6 1/2
Crux 1.7 beta 5 1/2
Amyan 1.0 beta 5
TSCP 1.5 3 1/2
Averno 0.27 3 1/2
Snail 1.03 2
Colchess 5.6 1 1/2
Freyr 0.960 1/2
Olithink is unknown in test! Why?
Crux did best than Amyan in it test, but lower in other previous
TSCP with the command -st 5 play for best
Freyr is a pity...

Please, send me the Partys of crux!
Use v01.8.
Cheers
WYx

[José Carlos]

Geschrieben von:/Posted by: José Carlos at 03 March 2000 17:48:12:
Als Antwort auf:/As an answer to: Re: MID-LOW Tournament geschrieben von:/posted by: Colin Frayn at 03 March 2000 12:11:42:

Olithink 2.2.1 6 1/2
Crux 1.7 beta 5 1/2
Amyan 1.0 beta 5
TSCP 1.5 3 1/2
Averno 0.27 3 1/2
Snail 1.03 2
Colchess 5.6 1 1/2
Freyr 0.960 1/2

That's the strangest tournament results I've seen for a long time.
I'm surprised Amyan and Olithink did so well. I don't know about Crux. Freyr is a strong program so why that came bottom I have no idea. Averno should have come higher. No comment about ColChess. TSCP never fails to amaze me. I've taken the code apart time and time again, and there's simply no way it should play as well as it does, within an order of magnitude. How it came so high, I have no idea. ColChess' record against it is usually 90% wins, though TSCP rarely uses a third of ColChess' thinking time.
Did you use ColChess' large opening book? I think that is going to be the default book for version 6.0 as lots of people just use the really small example book and don't realise that there's a 400k one with it...
Cheers,
Col

I guess that in so slow machine, and with so little time, the behaviour of these mid-low program become random, I mean, tactical mistakes are so big and so many, that you could replay the tourney and get the standings table upside down. I only can speak for my case, but I'm rather sure none of us have written our programs oriented to so slow machines.
José C.

[Colin Frayn]

Geschrieben von:/Posted by: Colin Frayn at 03 March 2000 21:12:29:
Als Antwort auf:/As an answer to: Re: MID-LOW Tournament geschrieben von:/posted by: José Carlos at 03 March 2000 17:48:12:

I guess that in so slow machine, and with so little time, the behaviour of these mid-low program become random, I mean, >tactical mistakes are so big and so many, that you could replay the tourney and get the standings table upside down.

Yes, I agree with you there. In fact I've never tested mine on slower hardware than my P3-450. I saw it running on a P-233 once and it was shockingly slow
I think that the other problem is that the higher-level code such as the hashtable doesn't really kick in until you start searching a lot of nodes or you get to an endgame with a lot of repetition.
As for null move pruning and other such methods - well I never really looked at them much as the idea of guesswork goes totally against my principles *grin* I suppose they could potentially work better on faster machines though as such methods really only pay off when you begin to spot quite deep tactical PVs and below a certain depth you're always fighting to avoid tactical mistakes instead. Also, at lower time controls the chances of you spotting something extremely cunning which doesn't look good until several ply is rather small.
Oh well - v6.0 of ColChess should sort this out I just played the development version against TSCP with the -st 5 option that Carlos Drake suggested, and it didn't even come close to losing any of the 9 games I played. Having said that, this was the case for v5.6 too.....
Cheers,
Colin

ColChess Homepage

[Andrei Fortuna]

Geschrieben von:/Posted by: Andrei Fortuna at 05 March 2000 05:48:43:
Als Antwort auf:/As an answer to: Re: MID-LOW Tournament geschrieben von:/posted by: Colin Frayn at 03 March 2000 12:11:42:

Olithink 2.2.1 6 1/2
Crux 1.7 beta 5 1/2
Amyan 1.0 beta 5
TSCP 1.5 3 1/2
Averno 0.27 3 1/2
Snail 1.03 2
Colchess 5.6 1 1/2
Freyr 0.960 1/2

That's the strangest tournament results I've seen for a long time.
I'm surprised Amyan and Olithink did so well. I don't know about Crux. Freyr is a strong program so why that came bottom I have no idea.

I do have an idea. In my latest 0.960 release of Freyr I forgot inside the makemove() a piece of debug code ( a memcpy(...,...,256)) which slowed down Freyr lots.
Returning to the results, it's indeed strange that Olithink scores so high, Freyr (latest without the ) can beat it pretty often (and on occasions I get a "second program rejects the move ..." from olithink under winboard for a perfectly valid move), while from 70 5:0 matches against Amyan Freyr wins 10, loses 54 and makes 6 draws.