Does anyone have an opinion on which of these programs is st

[Imran Hendley]

Geschrieben von: / Posted by: Imran Hendley at 23 April 2000 03:44:16:
Francesca 0.78, Arasan 5.3, Ant 4.16, Bringer 1.6, ZChess 1.51 and Dragon 3.11?

[Frank Quisinsky]

Geschrieben von: / Posted by: Frank Quisinsky at 23 April 2000 09:26:53:
Als Antwort auf: / As an answer to: Does anyone have an opinion on which of these programs is strongest? geschrieben von: / posted by: Imran Hendley at 23 April 2000 03:44:16:

Francesca 0.78, Arasan 5.3, Ant 4.16, Bringer 1.6, ZChess 1.51 and Dragon 3.11?

ZChess 1.51 is very strong but ZChess 1.51 has an mistake in the time managment. ZChess play the first 20 moves in 5 minutes (40 moves in 40 minutes).
But Frank Zibi know the problem (he found the bug in the time managment).
I think Dragon 3.11 play with 2250-2300 ELO, ZChess 1.51 Beta with 2350 ELO, Bringer with 2350 or a little bit stronger, Ant with 2300 (stronger with longer time controls), Arasan 5.3 with 2300-2350 ELO and Francesca with 2350-2400 ELO !
In this group furthermore InmiChess 3.03, Knightx 1.53, GnuChess 4.14, EXchess 3.11 and LambChop 7.1 !
The next group is SSEChess 2.02, Fortress 1.61, Amyan 1.01, Sjeng 6, La Dame Blanch 2.0c, ColChess 6.1 and the new Averno 0.28 with "?"
Regards
Frank

[Imran Hendley]

Geschrieben von: / Posted by: Imran Hendley at 23 April 2000 17:33:52:
Als Antwort auf: / As an answer to: Re: Does anyone have an opinion on which of these programs is strongest? geschrieben von: / posted by: Frank Quisinsky at 23 April 2000 09:26:53:
.

[Mogens Larsen]

Geschrieben von: / Posted by: Mogens Larsen at 23 April 2000 18:04:34:
Als Antwort auf: / As an answer to: Does anyone have an opinion on which of these programs is strongest? geschrieben von: / posted by: Imran Hendley at 23 April 2000 03:44:16:

Francesca 0.78, Arasan 5.3, Ant 4.16, Bringer 1.6, ZChess 1.51 and Dragon 3.11?

It's difficult to say as are fairly close in strength. I would eliminate Dragon and Arasan and make the choice between Ant, Bringer and Francesca. The strength of Zchess 1.51 is unknown to me. I like Bringer, so I'd probably choose it.
Best wishes...
Mogens
http://home1.stofanet.dk/Moq/

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 25 April 2000 08:25:35:
Als Antwort auf: / As an answer to: Does anyone have an opinion on which of these programs is strongest? geschrieben von: / posted by: Imran Hendley at 23 April 2000 03:44:16:

Francesca 0.78, Arasan 5.3, Ant 4.16, Bringer 1.6, ZChess 1.51 and Dragon 3.11?

Hi Imran,
ANT 4.16 and Bringer 1.6 have played relatively badly in my tournaments. Of the mentioned programs I think Francesca and Arasan are best (close), but I have never met ZChess 1.51.
In fact, ZChess 1.20 is strong enough, and I'd put it into the same category as Francesca and Arasan.
I hazard the opinion that best of all is Bringer 1.5 (yes, five!). I'm afraid 1.6 is considerably worse at all time controls.
Best regards,
Gábor

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 25 April 2000 08:42:22:
Als Antwort auf: / As an answer to: Re: Does anyone have an opinion on which of these programs is strongest? geschrieben von: / posted by: Gabor Szots at 25 April 2000 08:25:35:

Francesca 0.78, Arasan 5.3, Ant 4.16, Bringer 1.6, ZChess 1.51 and Dragon 3.11?

Hi Imran,
ANT 4.16 and Bringer 1.6 have played relatively badly in my tournaments. Of the mentioned programs I think Francesca and Arasan are best (close), but I have never met ZChess 1.51.
In fact, ZChess 1.20 is strong enough, and I'd put it into the same category as Francesca and Arasan.
I hazard the opinion that best of all is Bringer 1.5 (yes, five!). I'm afraid 1.6 is considerably worse at all time controls.

I am not so sure about Bringer 1.6 being weaker.
It seems to go in phases, and it appears to me that some programs "have bringer's number" and bringer "has other program's number."
In my calibration runs, Bringer 1.6 has done very respectably at game/1hr.
Fairly soon, there will be a large number of bringer games (probably 30-40 hours worth) to judge with.


My FTP site

[pete]

Geschrieben von: / Posted by: pete at 25 April 2000 20:01:28:
Als Antwort auf: / As an answer to: Re: Does anyone have an opinion on which of these programs is strongest? geschrieben von: / posted by: Dann Corbit at 25 April 2000 08:42:22:

Francesca 0.78, Arasan 5.3, Ant 4.16, Bringer 1.6, ZChess 1.51 and Dragon 3.11?

Hi Imran,
ANT 4.16 and Bringer 1.6 have played relatively badly in my tournaments. Of the mentioned programs I think Francesca and Arasan are best (close), but I have never met ZChess 1.51.
In fact, ZChess 1.20 is strong enough, and I'd put it into the same category as Francesca and Arasan.
I hazard the opinion that best of all is Bringer 1.5 (yes, five!). I'm afraid 1.6 is considerably worse at all time controls.

I am not so sure about Bringer 1.6 being weaker.
It seems to go in phases, and it appears to me that some programs "have bringer's number" and bringer "has other program's number."
In my calibration runs, Bringer 1.6 has done very respectably at game/1hr.
Fairly soon, there will be a large number of bringer games (probably 30-40 hours worth) to judge with.

I am _definitely_ sure that Bringer 1.6 is stronger in longer time controls than 1.5 was ; there are quite a few games in gambitmaster's history on ICC played with exactly this version . Also if you look at games played at longer time controls in user tournaments Bringer 1.6 does reasonably well .
I think some bad results by Bringer could have one of the following reasons :
1.) The default opening book is used.
This book really isn't very good . The "medium.bok" at Gerrit's web site is _much_ better and worth the time for download ; additionally it could be worth the time to tune this a little by removing some opining which seem to be bad for bringer with black like ben-oni or french but I expect already with medium.bok bringer gains about 50 points . Adding something like the 2600.zip from Dan's ftp-site helps improving the book further .
If someone is interested I could rebuild the book Bringer used in the final rounds of cct1 and send it to someone with an ftp-site .
2.) The default hash settings are used.
This will be a great problem at shorter time controls as the default values are very low .
Another minor thing is that it isn't always sure if tablebases are used : bringer uses them _very_ agressively and I expect a rather strong difference.
Another very slight trend seems to be that maybe Bringer is performing better on INTEL than on AMD ( so far only a guess ).

[pete]

Geschrieben von: / Posted by: pete at 25 April 2000 21:44:52:
Als Antwort auf: / As an answer to: About bringer geschrieben von: / posted by: pete at 25 April 2000 20:01:28:

This will be a great problem at shorter time controls as the default values are very low .

change shorter to longer please ; I wonder when I will post my first message without stuff like that
pete

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 25 April 2000 22:00:36:
Als Antwort auf: / As an answer to: Re: About bringer geschrieben von: / posted by: pete at 25 April 2000 21:44:52:

This will be a great problem at shorter time controls as the default values are very low .

change shorter to longer please ; I wonder when I will post my first message without stuff like that

I am dyslexic, which leads to all sorts of humorous gaffes that you could not imagine.
Once I went on and on about how a big win was missed because I read "0-1" as "1-0"



My FTP site

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 26 April 2000 10:23:47:
Als Antwort auf: / As an answer to: About bringer geschrieben von: / posted by: pete at 25 April 2000 20:01:28:

I hazard the opinion that best of all is Bringer 1.5 (yes, five!). I'm afraid 1.6 is considerably worse at all time controls.

I am not so sure about Bringer 1.6 being weaker.
It seems to go in phases, and it appears to me that some programs "have bringer's number" and bringer "has other program's number."
In my calibration runs, Bringer 1.6 has done very respectably at game/1hr.
Fairly soon, there will be a large number of bringer games (probably 30-40 hours worth) to judge with.

I am _definitely_ sure that Bringer 1.6 is stronger in longer time controls than 1.5 was ; there are quite a few games in gambitmaster's history on ICC played with exactly this version . Also if you look at games played at longer time controls in user tournaments Bringer 1.6 does reasonably well .
I think some bad results by Bringer could have one of the following reasons :
1.) The default opening book is used.
This book really isn't very good . The "medium.bok" at Gerrit's web site is _much_ better and worth the time for download ; additionally it could be worth the time to tune this a little by removing some opining which seem to be bad for bringer with black like ben-oni or french but I expect already with medium.bok bringer gains about 50 points . Adding something like the 2600.zip from Dan's ftp-site helps improving the book further .
If someone is interested I could rebuild the book Bringer used in the final rounds of cct1 and send it to someone with an ftp-site .
2.) The default hash settings are used.
This will be a great problem at shorter time controls as the default values are very low .
Another minor thing is that it isn't always sure if tablebases are used : bringer uses them _very_ agressively and I expect a rather strong difference.
Another very slight trend seems to be that maybe Bringer is performing better on INTEL than on AMD ( so far only a guess ).

I can refer only to the tournaments I have played with both Bringer versions. They both played using medium.bok and 16/16/2 MB hash setting. To judge the relative strength of 1.5 and 1.6 look at tournaments GS-1, GS-2 and GS-5 (currently still running):
GS-1 (15 min): v1.5 got 4.5 points out of 7, v1.6 got 2.5.
GS-2 (30 min): v1.5 got 3.5 points out of 7, v1.6 got 2.5.
GS-5 (40/40): v1.5 has 4.5 out of 8 games, v1.6 only 1.5 (!) out of 7. This makes an Elo difference of several hundred points!
I am quite aware that these are only 21-22 games each, but can it be a statistical freak?
Best regards,
Gábor

[pete]

Geschrieben von: / Posted by: pete at 26 April 2000 17:03:01:
Als Antwort auf: / As an answer to: Re: About bringer geschrieben von: / posted by: Gabor Szots at 26 April 2000 10:23:47:

I can refer only to the tournaments I have played with both Bringer versions. They both played using medium.bok and 16/16/2 MB hash setting. To judge the relative strength of 1.5 and 1.6 look at tournaments GS-1, GS-2 and GS-5 (currently still running):
GS-1 (15 min): v1.5 got 4.5 points out of 7, v1.6 got 2.5.
GS-2 (30 min): v1.5 got 3.5 points out of 7, v1.6 got 2.5.
GS-5 (40/40): v1.5 has 4.5 out of 8 games, v1.6 only 1.5 (!) out of 7. This makes an Elo difference of several hundred points!
I am quite aware that these are only 21-22 games each, but can it be a statistical freak?
Best regards,
Gábor

Indeed interesting tournament results : I never played winboard tournaments like this ( ponder off on one computer ) so I don't know if this might be a problem ( for example Gerrit changed time management in 1.6 to be more agressive and that could hurt without pondering ) .
There were also a few very good results of Bringer 1.6 posted but there were very few tournaments with both versions in it to compair directly .

[pete]

Geschrieben von: / Posted by: pete at 26 April 2000 18:22:37:
Als Antwort auf: / As an answer to: Re: About bringer geschrieben von: / posted by: pete at 26 April 2000 17:03:01:
I decided to invest some time to test this .
Theory : Bringer 1.6 is significantly stronger than Bringer 1.5 at longer time controls.
Bringer will play on a PIII500 against various opponents on a PII300 .
Time control will be 60/10 .
Both versions will use the same book ( medium.bok ) as you used and both versions will be given the same amount for Hash tables .
It will take some time : I will post the results when I am ready .
greetings.
pete

[pete]

Geschrieben von: / Posted by: pete at 26 April 2000 21:46:31:
Als Antwort auf: / As an answer to: Re: About bringer / Test geschrieben von: / posted by: pete at 26 April 2000 18:22:37:

I decided to invest some time to test this .
Theory : Bringer 1.6 is significantly stronger than Bringer 1.5 at longer time controls.
Bringer will play on a PIII500 against various opponents on a PII300 .
Time control will be 60/10 .
Both versions will use the same book ( medium.bok ) as you used and both versions will be given the same amount for Hash tables .
It will take some time : I will post the results when I am ready .
greetings.
pete

The 1st game showed one of the weakest Bringer games I ever have seen ...
[Event "60/10"]
[Site "unbekannt"]
[Date "2000.04.26"]
[Round "1"]
[White "Bringer 1.6(PIII500)"]
[Black "Virtual Chess II ( PII300)"]
[Result "Schwarz gewinnt"]
1. e4 { 0.00 0:00:00.0 } e5 2. Nf3 { 0.00 0:00:00.0 } Nc6
3. Bb5 { 0.00 0:00:00.0 } a6 4. Ba4 { 0.00 0:00:00.0 } Nf6
5. O-O { 0.00 0:00:00.0 } Be7 6. Qe2 { 0.00 0:00:00.0 } b5
7. Bb3 { 0.00 0:00:00.0 } O-O 8. c3 { 0.00 0:00:00.0 } d6
9. Rd1 { 0.00 0:00:00.0 } Na5 10. Bc2 { 0.00 0:00:00.0 } c5
11. d4 { 0.00 0:00:00.0 } Qc7 12. d5 { 0.00 0:00:00.0 } h6
13. Nbd2 { 0.10/11 0:02:03.1 } Bg4 14. b3 { 0.18/10 0:01:29.4 }
Bd7 15. Re1 { 0.18/10 0:01:19.7 (PB: 0:01:19.7 ) } Rfc8 16. Rd1 { 0.17/10 0:01:35.8 }
Nb7 17. c4 { 0.17/11 0:01:31.3 } b4 18. a3 { 0.25/12 0:02:08.1 }
a5 19. Nf1 { 0.17/10 0:01:14.7 (PB: 0:01:14.6 ) } a4 20. bxa4 { 0.17/10 0:01:07.5 (PB: 0:01:07.4 ) }
Bxa4 21. Bxa4 { 0.19/10 0:01:05.2 (PB: 0:01:05.1 ) } Rxa4 22. Bb2 { 0.10/10 0:01:36.5 (PB: 0:01:36.5 ) }
Rca8 23. axb4 { 0.10/10 0:01:06.7 (PB: 0:01:06.6 ) } cxb4 24. Rxa4 { 0.03/10 0:01:22.0 (PB: 0:01:22.0 ) }
Rxa4 25. Ng3 { 2.46/11 0:01:32.7 (PB: 0:01:32.6 ) } Nc5 26. Kh1 {-0.03/10 0:01:11.2 }
Ra2 27. Rd2 {-0.81/11 0:01:06.7 } Na4 28. Ng1 {-0.98/11 0:01:45.6 (PB: 0:00:09.0 ) }
b3 29. Bc1 {-1.01/11 0:01:39.0 } Rxd2 30. Bxd2 {-1.50/12 0:09:50.8 }
b2 31. Qf1 {-2.30/11 0:01:42.8 } Qc5 32. N1e2 {-2.31/9 0:00:48.0 }
Ng4 33. Be3 {-5.19/10 0:00:46.6 } Nxe3 34. fxe3 {-8.63/13 0:01:30.3 }
Qa3 35. c5 {-9.30/12 0:01:35.9 } Qa1 36. cxd6 {-9.64/12 0:01:43.2 }
Schwarz gewinnt
Virtual Chess _can_ play like a monster but a bad start this really is ...
26. Kh1 , 28. Ng1 : oh man !
Bringer is out of practise it seems
We will see but this really was too much for one evening for me
pete

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 27 April 2000 08:07:55:
Als Antwort auf: / As an answer to: Re: About bringer / Test geschrieben von: / posted by: pete at 26 April 2000 18:22:37:

I decided to invest some time to test this .
Theory : Bringer 1.6 is significantly stronger than Bringer 1.5 at longer time controls.
Bringer will play on a PIII500 against various opponents on a PII300 .
Time control will be 60/10 .
Both versions will use the same book ( medium.bok ) as you used and both versions will be given the same amount for Hash tables .
It will take some time : I will post the results when I am ready .
greetings.
pete

Hi,
I like this idea. Using ponder on may produce quite different results. However, is this time control not too short for Bringer? Remember your theory!
Best regards,
Gábor

[pete]

Geschrieben von: / Posted by: pete at 27 April 2000 13:26:10:
Als Antwort auf: / As an answer to: Re: About bringer / Test geschrieben von: / posted by: Gabor Szots at 27 April 2000 08:07:55:

I decided to invest some time to test this .
Theory : Bringer 1.6 is significantly stronger than Bringer 1.5 at longer time controls.
Bringer will play on a PIII500 against various opponents on a PII300 .
Time control will be 60/10 .
Both versions will use the same book ( medium.bok ) as you used and both versions will be given the same amount for Hash tables .
It will take some time : I will post the results when I am ready .
greetings.
pete

Hi,
I like this idea. Using ponder on may produce quite different results. However, is this time control not too short for Bringer? Remember your theory!
Best regards,
Gábor

60/10 is Fisher Time : Game in 60 with 10 seconds increment per move.

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 27 April 2000 14:50:54:
Als Antwort auf: / As an answer to: Re: About bringer / Test geschrieben von: / posted by: pete at 27 April 2000 13:26:10:

I decided to invest some time to test this .
Theory : Bringer 1.6 is significantly stronger than Bringer 1.5 at longer time controls.
Bringer will play on a PIII500 against various opponents on a PII300 .
Time control will be 60/10 .
Both versions will use the same book ( medium.bok ) as you used and both versions will be given the same amount for Hash tables .
It will take some time : I will post the results when I am ready .
greetings.
pete

Hi,
I like this idea. Using ponder on may produce quite different results. However, is this time control not too short for Bringer? Remember your theory!
Best regards,
Gábor

60/10 is Fisher Time : Game in 60 with 10 seconds increment per move.

OK. I had the impression it was 60 moves in 10 minutes.