Le Système du Suisse : Round 4 (40'/40 moves)

[Claude Dubois]

[Alejandro Dubrovsky]

Geschrieben von:/Posted by: Alejandro Dubrovsky at 24 March 2004 10:02:27:
Als Antwort auf:/In reply to: Le Système du Suisse : Round 4 (40'/40 moves) geschrieben von:/posted by: Claude Dubois at 24 March 2004 08:29:51:

If i understand correctly, you mention that small potato resigned in the game but you overrode the result. (Thanks) I don't think it's an evaluation bug, but a bad interaction with WB2UCI (its seeing mate at that point, no need for proper evaluation). Is it possible that the interface is resigning for the engine when the engine reports a very bad evaluation score? (Potato always reports score for the side to play, not for white, and some interfaces expect white score)



SmallPotato homepage

[Claude Dubois]

Geschrieben von:/Posted by: Claude Dubois at 24 March 2004 14:06:53:
Als Antwort auf:/In reply to: Re: Le Système du Suisse : Round 4 (40'/40 moves) geschrieben von:/posted by: Alejandro Dubrovsky at 24 March 2004 10:02:27:

If i understand correctly, you mention that small potato resigned in the game but you overrode the result. (Thanks) I don't think it's an evaluation bug, but a bad interaction with WB2UCI (its seeing mate at that point, no need for proper evaluation). Is it possible that the interface is resigning for the engine when the engine reports a very bad evaluation score? (Potato always reports score for the side to play, not for white, and some interfaces expect white score)

Hi Alejandro,
Thanks for your interest.
You have good undersand my french.
I have tried in my test the parameter whitescore=true with Small potato but I have the feeling that it's not the problem.
The problem is happened when SmallPotato has a big advantage and he doesn't know make evaluations.
I have sent to you the log file of the game and 6 games 3'/40 moves with the file wb2uci.log. The 3rd game has a similar problem.
You can always see the evaluation after the move in the games.
For example, in the game of the 4th round the 41th move Small potato evalue right with -20.60 and during all the game the informations to the GUI were good and with the informationS as whitescore. But in the 42th move, the evaluation is noting 272.03. This information disturb all the game.
I am going to let the logfile of wb2uci.exe next rounds.
Best wishes
Claude



Le Système du Suisse

[Alejandro Dubrovsky]

Geschrieben von:/Posted by: Alejandro Dubrovsky at 24 March 2004 16:15:33:
Als Antwort auf:/In reply to: Re: Le Système du Suisse : Round 4 (40'/40 moves) geschrieben von:/posted by: Claude Dubois at 24 March 2004 14:06:53:

You have good undersand my french.
I have tried in my test the parameter whitescore=true with Small potato but I have the feeling that it's not the problem.
The problem is happened when SmallPotato has a big advantage and he doesn't know make evaluations.
I have sent to you the log file of the game and 6 games 3'/40 moves with the file wb2uci.log. The 3rd game has a similar problem.
You can always see the evaluation after the move in the games.
For example, in the game of the 4th round the 41th move Small potato evalue right with -20.60 and during all the game the informations to the GUI were good and with the informationS as whitescore. But in the 42th move, the evaluation is noting 272.03. This information disturb all the game.
I am going to let the logfile of wb2uci.exe next rounds.
Best wishes
Claude

Thanks for the log files, i had a look. Exactly the same problem, but it's not a problem with potato. Potato never resigns in that file (otherwise it says "Black resigns" or "White resigns"). Its evaluation is showing 999995 (which means mate) in both cases, which is correct. Wb2UCI is getting confused by this and just quitting. Maybe it doesn't like scores beyond some range. I don't have access to a windows machine so its hard for me to test wb2uci. Anyone else have any experience with this?
Potato would never show 272.03. It's maximum normal scores would be around 18 or so, and by then, it's most likely showing mate (which is around 999990). This is some interpretation by the interface.
Thanks for the info. I'll have to learn how not to confuse wb2uci



SmallPotato homepage

[Claude Dubois]

Geschrieben von:/Posted by: Claude Dubois at 24 March 2004 17:08:38:
Als Antwort auf:/In reply to: Re: Le Système du Suisse : Round 4 (40'/40 moves) geschrieben von:/posted by: Alejandro Dubrovsky at 24 March 2004 16:15:33:

You have good undersand my french.
I have tried in my test the parameter whitescore=true with Small potato but I have the feeling that it's not the problem.
The problem is happened when SmallPotato has a big advantage and he doesn't know make evaluations.
I have sent to you the log file of the game and 6 games 3'/40 moves with the file wb2uci.log. The 3rd game has a similar problem.
You can always see the evaluation after the move in the games.
For example, in the game of the 4th round the 41th move Small potato evalue right with -20.60 and during all the game the informations to the GUI were good and with the informationS as whitescore. But in the 42th move, the evaluation is noting 272.03. This information disturb all the game.
I am going to let the logfile of wb2uci.exe next rounds.
Best wishes
Claude

Thanks for the log files, i had a look. Exactly the same problem, but it's not a problem with potato. Potato never resigns in that file (otherwise it says "Black resigns" or "White resigns"). Its evaluation is showing 999995 (which means mate) in both cases, which is correct. Wb2UCI is getting confused by this and just quitting. Maybe it doesn't like scores beyond some range. I don't have access to a windows machine so its hard for me to test wb2uci. Anyone else have any experience with this?
Potato would never show 272.03. It's maximum normal scores would be around 18 or so, and by then, it's most likely showing mate (which is around 999990). This is some interpretation by the interface.
Thanks for the info. I'll have to learn how not to confuse wb2uci

Hello Alejandro,
Don't worry about it. I have made the error.
I have let played SmallPotato with the wb2uci's version of the 22.07.2003 and Odd Gunnar Malin has made a new version dated from the 06.01.2004.
In this version, I have read that Odd Gunnar has implanted the parameter MatScore that check the value sent bye the engine when the mat is detected.
I think that it is the solution to your problem. I am going to continue test with the new wb2uci.exe and give feedback to you.
When we have a lot of engines to manage I wanted to wait for the end of the tournament to check the new version. Only updates and new engines are playing with the new wb2uci.exe.
Thanks for all.
Best wishes
Claude



Le Système du Suisse

[Alejandro Dubrovsky]

Geschrieben von:/Posted by: Alejandro Dubrovsky at 24 March 2004 18:42:48:
Als Antwort auf:/In reply to: Re: Le Système du Suisse : Round 4 (40'/40 moves) geschrieben von:/posted by: Claude Dubois at 24 March 2004 17:08:38:

Hello Alejandro,
Don't worry about it. I have made the error.
I have let played SmallPotato with the wb2uci's version of the 22.07.2003 and Odd Gunnar Malin has made a new version dated from the 06.01.2004.
In this version, I have read that Odd Gunnar has implanted the parameter MatScore that check the value sent bye the engine when the mat is detected.
I think that it is the solution to your problem. I am going to continue test with the new wb2uci.exe and give feedback to you.
When we have a lot of engines to manage I wanted to wait for the end of the tournament to check the new version. Only updates and new engines are playing with the new wb2uci.exe.
Thanks for all.

Thanks for the update. Whenever i get access to a windows machine, i'll do some wb2uci testing.




SmallPotato homepage