Cerebro 1.25 gauntlet---Frenzee 151 note

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 08 April 2004 10:40:01:

Hi ,
a 480 4+2 Blitz games gauntlet over all Nunn positions on Athlon 2600+ with the following results and an improvement for Cerebro 1.25 over 1.23 by 12 points.


Cerebro 1.25   - Snitch 0.3.47      18.5 - 21.5
Cerebro 1.25   - Amateur 2.78       18.5 - 21.5
Cerebro 1.25   - Queen 2.42a        19.5 - 20.5
Cerebro 1.25   - Averno 0.70        17.5 - 22.5
Cerebro 1.25   - Chezzz 1.0.3       13.0 - 27.0
Cerebro 1.25   - Ufim 4.04          21.5 - 18.5
Cerebro 1.25   - Trace 1.27         20.5 - 19.5
Cerebro 1.25   - Chispa v4.041      20.5 - 19.5
Cerebro 1.25   - Booot 3.1          20.0 - 20.0
Cerebro 1.25   - Fruit 1.0          16.0 - 24.0
Cerebro 1.25   - Movei 00_8_178     10.5 - 29.5
Cerebro 1.25   - Frenzee 151        08.5 - 31.5



Running the last gauntlets I noticed that there were a lot of wrong result outputs for Frenzee 151. So I manually checked all games from Nunn B tournament and match versus AnMon 5.30 that I could not watch on the other computer. And there were indeed many Frenzee wins where it outputs a loss. I know that this happens with some engines and can be usually avoided by adding WhiteScore=true and/or InitString=set resign 0 in the Wb2UCI file. But I was not aware of that problem in case of Frenzee, as the Frenzee UCI versions before did not give wrong results.
With all those Frenzee 151 results corrected there is a huge improvement over version 145. The match against AnMon was even won by 22:18 and in the Nunn B tournament Frenzee is no longer last, but on rank 12 in the final total score. Rating lists and crosstable for Total Score Nunn B are updated and corrected.
The tables for each Nunn position in Tournament B will be corrected over the following days.
http://www.husvankempen.de/nunn/
http://www.husvankempen.de/nunn/tour/index.htm
Best Regards
Heinz

[Claude Dubois]

Geschrieben von:/Posted by: Claude Dubois at 08 April 2004 11:33:32:
Als Antwort auf:/In reply to: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Heinz van Kempen at 08 April 2004 10:40:01:

Hi ,
a 480 4+2 Blitz games gauntlet over all Nunn positions on Athlon 2600+ with the following results and an improvement for Cerebro 1.25 over 1.23 by 12 points.



Running the last gauntlets I noticed that there were a lot of wrong result outputs for Frenzee 151. So I manually checked all games from Nunn B tournament and match versus AnMon 5.30 that I could not watch on the other computer. And there were indeed many Frenzee wins where it outputs a loss. I know that this happens with some engines and can be usually avoided by adding WhiteScore=true and/or InitString=set resign 0 in the Wb2UCI file. But I was not aware of that problem in case of Frenzee, as the Frenzee UCI versions before did not give wrong results.
With all those Frenzee 151 results corrected there is a huge improvement over version 145. The match against AnMon was even won by 22:18 and in the Nunn B tournament Frenzee is no longer last, but on rank 12 in the final total score. Rating lists and crosstable for Total Score Nunn B are updated and corrected.
The tables for each Nunn position in Tournament B will be corrected over the following days.
http://www.husvankempen.de/nunn/
http://www.husvankempen.de/nunn/tour/index.htm
Best Regards
Heinz

>Cerebro 1.25   - Snitch 0.3.47      18.5 - 21.5
>Cerebro 1.25   - Amateur 2.78       18.5 - 21.5
>Cerebro 1.25   - Queen 2.42a        19.5 - 20.5
>Cerebro 1.25   - Averno 0.70        17.5 - 22.5
>Cerebro 1.25   - Chezzz 1.0.3       13.0 - 27.0
>Cerebro 1.25   - Ufim 4.04          21.5 - 18.5
>Cerebro 1.25   - Trace 1.27         20.5 - 19.5
>Cerebro 1.25   - Chispa v4.041      20.5 - 19.5
>Cerebro 1.25   - Booot 3.1          20.0 - 20.0
>Cerebro 1.25   - Fruit 1.0          16.0 - 24.0
>Cerebro 1.25   - Movei 00_8_178     10.5 - 29.5
>Cerebro 1.25   - Frenzee 151        08.5 - 31.5
>
>
Hello Heinz,
Have a try with it
[ENGINE]
Name=Frenzee 151 W2U
Author=Sune Fisher, Danemark
Filename=Wb2Uci.exe
[OPTIONS]
Program=frenzee.exe
WhiteScore=true
MateScore=299987
Don't forget to take the last version of the march 25th 2004 from wb2uci.exe
Best wishes
Claude
PS SmallPotato has also the same problem and you also can try MateScore=9999800
DChess 1.02 is the ideal sparringspartner for the test


Le Système du Suisse

[Sune Fischer]

Geschrieben von:/Posted by: Sune Fischer at 08 April 2004 11:44:02:
Als Antwort auf:/In reply to: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Heinz van Kempen at 08 April 2004 10:40:01:

Hi ,
a 480 4+2 Blitz games gauntlet over all Nunn positions on Athlon 2600+ with the following results and an improvement for Cerebro 1.25 over 1.23 by 12 points.



Running the last gauntlets I noticed that there were a lot of wrong result outputs for Frenzee 151. So I manually checked all games from Nunn B tournament and match versus AnMon 5.30 that I could not watch on the other computer. And there were indeed many Frenzee wins where it outputs a loss. I know that this happens with some engines and can be usually avoided by adding WhiteScore=true and/or InitString=set resign 0 in the Wb2UCI file. But I was not aware of that problem in case of Frenzee, as the Frenzee UCI versions before did not give wrong results.
With all those Frenzee 151 results corrected there is a huge improvement over version 145. The match against AnMon was even won by 22:18 and in the Nunn B tournament Frenzee is no longer last, but on rank 12 in the final total score. Rating lists and crosstable for Total Score Nunn B are updated and corrected.
The tables for each Nunn position in Tournament B will be corrected over the following days.
http://www.husvankempen.de/nunn/
http://www.husvankempen.de/nunn/tour/index.htm
Best Regards
Heinz

>Cerebro 1.25   - Snitch 0.3.47      18.5 - 21.5
>Cerebro 1.25   - Amateur 2.78       18.5 - 21.5
>Cerebro 1.25   - Queen 2.42a        19.5 - 20.5
>Cerebro 1.25   - Averno 0.70        17.5 - 22.5
>Cerebro 1.25   - Chezzz 1.0.3       13.0 - 27.0
>Cerebro 1.25   - Ufim 4.04          21.5 - 18.5
>Cerebro 1.25   - Trace 1.27         20.5 - 19.5
>Cerebro 1.25   - Chispa v4.041      20.5 - 19.5
>Cerebro 1.25   - Booot 3.1          20.0 - 20.0
>Cerebro 1.25   - Fruit 1.0          16.0 - 24.0
>Cerebro 1.25   - Movei 00_8_178     10.5 - 29.5
>Cerebro 1.25   - Frenzee 151        08.5 - 31.5
>
>
Hi Heinz,
This is great news indeed, almost too good to be true
It didn't occur to me something was wrong, I simply had very little idea how good it would be compared to the other improved engines.
I still can't help but to feel a bit unhappy with all the problems in this version. I hope I've seen the worst of it by now.
Happy Easter
-S.

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 08 April 2004 11:45:11:
Als Antwort auf:/In reply to: Re: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Claude Dubois at 08 April 2004 11:33:32:

Hello Heinz,
Have a try with it
[ENGINE]
Name=Frenzee 151 W2U
Author=Sune Fisher, Danemark
Filename=Wb2Uci.exe
[OPTIONS]
Program=frenzee.exe
WhiteScore=true
MateScore=299987
Don't forget to take the last version of the march 25th 2004 from wb2uci.exe
Best wishes
Claude
PS SmallPotato has also the same problem and you also can try MateScore=9999800
DChess 1.02 is the ideal sparringspartner for the test

Hello Claude ,
thanks a lot. This is a very useful hint. You can imagine how "funny" it is to check more than 700 games manually for correct results, as the utilities for that seem to work only for Winboard pgn-files.
Best Regards
Heinz

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 08 April 2004 11:49:34:
Als Antwort auf:/In reply to: Re: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Sune Fischer at 08 April 2004 11:44:02:

Hi Heinz,
This is great news indeed, almost too good to be true
It didn't occur to me something was wrong, I simply had very little idea how good it would be compared to the other improved engines.
I still can't help but to feel a bit unhappy with all the problems in this version. I hope I've seen the worst of it by now.
Happy Easter
-S.

Hi Sune,
nonetheless you have to admit that Frenzee is on the way to the top now. This is really a huge improvement. As soon as you have something even better, please send it to me, as soon a new Blitz tournaments is coming up, where Frenzee would fit in as one of the favourites.
Best Regards
Heinz

[Sune Fischer]

Geschrieben von:/Posted by: Sune Fischer at 08 April 2004 11:49:56:
Als Antwort auf:/In reply to: Re: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Claude Dubois at 08 April 2004 11:33:32:

Hello Heinz,
Have a try with it
[ENGINE]
Name=Frenzee 151 W2U
Author=Sune Fisher, Danemark
Filename=Wb2Uci.exe
[OPTIONS]
Program=frenzee.exe
WhiteScore=true
MateScore=299987

This score would indicate white mates in 7.
300000 is checkmate on the board but 299999 is the highest it can go
on the output line, eg.
pv: 57. Qxg7++ score: 299999 (mate in 1)
-S.

[Sune Fischer]

Geschrieben von:/Posted by: Sune Fischer at 08 April 2004 11:57:45:
Als Antwort auf:/In reply to: Re: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Heinz van Kempen at 08 April 2004 11:49:34:

Hi Heinz,
This is great news indeed, almost too good to be true
It didn't occur to me something was wrong, I simply had very little idea how good it would be compared to the other improved engines.
I still can't help but to feel a bit unhappy with all the problems in this version. I hope I've seen the worst of it by now.
Happy Easter
-S.

Hi Sune,
nonetheless you have to admit that Frenzee is on the way to the top now. This is really a huge improvement. As soon as you have something even better, please send it to me, as soon a new Blitz tournaments is coming up, where Frenzee would fit in as one of the favourites.
Best Regards
Heinz

I should like nothing better than to send you an improved version,
unfortunately my tedious systematical way of working isn't quite as speedy as
that of certain other leaps and bounds improving authors.

-S.

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 08 April 2004 12:14:35:
Als Antwort auf:/In reply to: Re: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Sune Fischer at 08 April 2004 11:57:45:

I should like nothing better than to send you an improved version,
unfortunately my tedious systematical way of working isn't quite as speedy as
that of certain other leaps and bounds improving authors.

-S.

Hi Sune,
take your time to have another improvement. Next to come in a few days are first results from Frenzee in Nunn Active C with 20 min. + 10 sec.
Best Regards
Heinz

[Claude Dubois]

Geschrieben von:/Posted by: Claude Dubois at 08 April 2004 14:04:18:
Als Antwort auf:/In reply to: Re: Cerebro 1.25 gauntlet---Frenzee 151 note geschrieben von:/posted by: Sune Fischer at 08 April 2004 11:49:56:

Hello Heinz,
Have a try with it
[ENGINE]
Name=Frenzee 151 W2U
Author=Sune Fisher, Danemark
Filename=Wb2Uci.exe
[OPTIONS]
Program=frenzee.exe
WhiteScore=true
MateScore=299987

This score would indicate white mates in 7.
300000 is checkmate on the board but 299999 is the highest it can go
on the output line, eg.
pv: 57. Qxg7++ score: 299999 (mate in 1)
-S.

Hello Sune,
My dread was to never reach the value 299999 in the game and that Frenzee give up before that. It was the idea to put a more little value. But I have made test with 299999 and it seems ok.
Thanks for Frenzee, it is a very interesting engine who has made a lot of improvement last time.
Best wishes
Claude



Le Système du Suisse