My tournament

[Pete Galati]

Geschrieben von:/Posted by: Pete Galati am 19. Dezember 1999 07:06:52:
At the moment Comet is playing matches. The link that I provided earlier to view the games was not updating itself with the changes, hopefully this will work though. If you're interested, I'll be uploading the text file probably after every match, and if I remember I'll put the pgnres at the top of the file, Winboard will ignore that (pgnres) if you save the file as a pgn file for viwing with Winboard. Again, it's now a text file, not an html page, so use your browser's "back" button too return to the forum.
Pete

Results from file knightx.pgn:
No. Name Win Draw Loss Unf. Score Games %
----------------------------------------------------------------------
1 TCB +41 =8 -11 *0 45.0 60 75.0%
2 Crafty +39 =11 -10 *0 44.5 60 74.2%
3 Knightx +26 =8 -26 *1 30.0 60 50.0%
4 Comet-wb +20 =7 -9 *1 23.5 36 65.3%
5 Phalanx +11 =4 -9 *0 13.0 24 54.2%
6 Gromit20 +6 =4 -14 *0 8.0 24 33.3%
7 GNUchess +5 =3 -16 *0 6.5 24 27.1%
8 Dragon311 +3 =1 -14 *0 3.5 18 19.4%
9 Fortress +2 =2 -14 *0 3.0 18 16.7%
10 GNUches5 +2 =1 -15 *0 2.5 18 13.9%
11 TSCP +0 =1 -17 *0 0.5 18 2.8%
Total Games: 181
White Wins: 90 (49.7%)
Black Wins: 65 (35.9%)
Draws: 25 (13.8%)
Unfinished: 1 (0.6%)


--To view the pgn file--

[Pete Galati]

Geschrieben von:/Posted by: Pete Galati am 19. Dezember 1999 23:29:10:
Als Antwort auf:/As an answer to: My tournament geschrieben von:/posted by: Pete Galati am 19. Dezember 1999 07:06:52:
Comet is finished playing, Phalanx is next.
Again, to view the pgn file with the below link, it's now a text file, not an html page, so use your browser's "back" button or Alt(left arrow) too return to the forum.
Pete

Results from file knightx.pgn:
No. Name Win Draw Loss Unf. Score Games %
----------------------------------------------------------------------
1 Comet-wb +42 =7 -11 *0 45.5 60 75.8%
2 TCB +41 =8 -11 *0 45.0 60 75.0%
3 Crafty +39 =11 -10 *0 44.5 60 74.2%
4 Knightx +26 =8 -26 *0 30.0 60 50.0%
5 Phalanx +11 =4 -9 *0 13.0 24 54.2%
6 Gromit20 +6 =4 -14 *0 8.0 24 33.3%
7 GNUchess +5 =3 -16 *0 6.5 24 27.1%
8 Dragon311 +4 =1 -19 *0 4.5 24 18.8%
9 GNUches5 +3 =1 -20 *0 3.5 24 14.6%
10 Fortress +2 =2 -20 *0 3.0 24 12.5%
11 TSCP +0 =1 -23 *0 0.5 24 2.1%
Total Games: 204
White Wins: 104 (51.0%)
Black Wins: 75 (36.8%)
Draws: 25 (12.3%)
Unfinished: 0 (0.0%)

View the games, pgn file

[Andrew Tanner]

Geschrieben von:/Posted by: Andrew Tanner am 20. Dezember 1999 11:24:12:
Als Antwort auf:/As an answer to: Re: Comet has finished playing geschrieben von:/posted by: Pete Galati am 19. Dezember 1999 23:29:10:
Hi Pete,
Just curious to see what the tournament results of Gromit might be if these settings were to be used in the eval.cfg file:
TRANSPOSITIONSIZE = 14
PAWNHASHSIZE = 14
EXCHANGEBONUS = 788
CFAKTORP = 72 ;22 x/32 * posbonustab pawn
CFAKTORN = 23 ;21 x/32 * posbonustab knight
CFAKTORB = 23 ;21 x/32 * posbonustab bishop
CFAKTORR = 18 ;16 x/32 * posbonustab rook
CFAKTORQ = 55 ;55 x/32 * posbonustab queen
CFAKTORK = 17 ;15 x/32 * posbonustab king
OPPOSITION = 7 ; 4 king in opposition in endgame

[Pete Galati]

Geschrieben von:/Posted by: Pete Galati am 20. Dezember 1999 17:20:06:
Als Antwort auf:/As an answer to: Would you add these special Gromit settings to your next tournament? geschrieben von:/posted by: Andrew Tanner am 20. Dezember 1999 11:24:12:

Hi Pete,
Just curious to see what the tournament results of Gromit might be if these settings were to be used in the eval.cfg file:
TRANSPOSITIONSIZE = 14
PAWNHASHSIZE = 14
EXCHANGEBONUS = 788
CFAKTORP = 72 ;22 x/32 * posbonustab pawn
CFAKTORN = 23 ;21 x/32 * posbonustab knight
CFAKTORB = 23 ;21 x/32 * posbonustab bishop
CFAKTORR = 18 ;16 x/32 * posbonustab rook
CFAKTORQ = 55 ;55 x/32 * posbonustab queen
CFAKTORK = 17 ;15 x/32 * posbonustab king
OPPOSITION = 7 ; 4 king in opposition in endgame

That might be too much for my computer, I have a rather humble amount of ram (40 mb), Gromit hasn't done real well in my tournament, at the moment I have:
TRANSPOSITIONSIZE = 12
PAWNHASHSIZE = 12
MATERIALHASHSIZE = 8
EVALHASHSIZE = 12
Pretty much everything else is standard, I don't recall what I might have changed above.
I hadn't meant to start a tournament, it was an accident that it got going so I don't have any real plans to start another one, and I can't really change the settings while in mid tournament. This is how things stand now this morning when I woke up:

Results from file knightx.pgn:
No. Name Win Draw Loss Unf. Score Games %
----------------------------------------------------------------------
1 Comet-wb +42 =7 -11 *0 45.5 60 75.8%
2 TCB +41 =8 -11 *0 45.0 60 75.0%
3 Crafty +39 =11 -10 *0 44.5 60 74.2%
4 Phalanx +28 =6 -14 *0 31.0 48 64.6%
5 Knightx +26 =8 -26 *0 30.0 60 50.0%
6 Gromit20 +9 =5 -16 *0 11.5 30 38.3%
7 GNUchess +6 =3 -21 *0 7.5 30 25.0%
8 Dragon311 +4 =2 -24 *0 5.0 30 16.7%
9 GNUches5 +4 =1 -25 *0 4.5 30 15.0%
10 Fortress +2 =2 -20 *0 3.0 24 12.5%
11 TSCP +0 =1 -23 *0 0.5 24 2.1%
Total Games: 228
White Wins: 116 (50.9%)
Black Wins: 85 (37.3%)
Draws: 27 (11.8%)
Unfinished: 0 (0.0%)

I think Gromit should be doing better than 50%, Phalanx still has to play 6 game matches against Fortress and TSCP, and then it'll be Gromit playing 6 game matches against all the programs below it, I have no idea how it will finish, I didn't expect Comet to be rather firmly in the lead.
Pete


to view the games in a text file

[Volker Pittlik]

Geschrieben von:/Posted by: Volker Pittlik am 20. Dezember 1999 20:06:33:
Als Antwort auf:/As an answer to: Re: Would you add these special Gromit settings to your next tournament? geschrieben von:/posted by: Pete Galati am 20. Dezember 1999 17:20:06:
Hi Pete,
...

That might be too much for my computer, I have a rather humble amount of ram (40 mb), Gromit hasn't done real well in my tournament, at the moment I have:
TRANSPOSITIONSIZE = 12
PAWNHASHSIZE = 12
MATERIALHASHSIZE = 8
EVALHASHSIZE = 12
Pretty much everything else is standard, I don't recall what I might have changed above.

That seems to be rather small. Evaluationhashtable isn't in use. As you have 40MB of RAM I suggest to increase the size of transpositiontable.
From my EVAL.CFG:
; The size of the transpositiontable where positions are stored
TRANSPOSITIONSIZE = 21 ; 2^n entries (n==TRANSPOSITIONSIZE)
; NEW: one entry now needs 16 byte (was 12 byte)
;22=64MB, 21=32MB, 20=16MB, 19=8MB, 18=4MB, 17=2MB, 16=1MB, 15=524K, 14=256K
; The size of the pawnhashtable is far less important than the
; size of the transposition-table. Values above 18 (2^18 entries)
; do not make much sense.
PAWNHASHSIZE = 18 ; 2^n entries (n==PAWHNHASHSIZE)
; one entry needs 8 Byte
;17=1MB, 16=524K, 15=256K, 14=128K, 13=64K, 12=32K
; The size of the materialhashtable
MATERIALHASHSIZE = 15 ; 2^n entries (n==MATERIALHASHSIZE)
; one entry needs 8 Byte
;12=32K, 11=16K, 10=8K, 9=4K, 8=2k
; The size of the evaluationhashtable
; currently not used!
EVALHASHSIZE = 1 ; 2^n entries (n==EVALHASHSIZE)
; one entry needs 8 Byte
Best wishes
Volker

[Pete Galati]

Geschrieben von:/Posted by: Pete Galati am 20. Dezember 1999 20:49:36:
Als Antwort auf:/As an answer to: Re: Would you add these special Gromit settings to your next tournament? geschrieben von:/posted by: Volker Pittlik am 20. Dezember 1999 20:06:33:

Hi Pete,
...

That might be too much for my computer, I have a rather humble amount of ram (40 mb), Gromit hasn't done real well in my tournament, at the moment I have:
TRANSPOSITIONSIZE = 12
PAWNHASHSIZE = 12
MATERIALHASHSIZE = 8
EVALHASHSIZE = 12
Pretty much everything else is standard, I don't recall what I might have changed above.

That seems to be rather small. Evaluationhashtable isn't in use. As you have 40MB of RAM I suggest to increase the size of transpositiontable.
From my EVAL.CFG:
; The size of the transpositiontable where positions are stored
TRANSPOSITIONSIZE = 21 ; 2^n entries (n==TRANSPOSITIONSIZE)
; NEW: one entry now needs 16 byte (was 12 byte)
;22=64MB, 21=32MB, 20=16MB, 19=8MB, 18=4MB, 17=2MB, 16=1MB, 15=524K, 14=256K
; The size of the pawnhashtable is far less important than the
; size of the transposition-table. Values above 18 (2^18 entries)
; do not make much sense.
PAWNHASHSIZE = 18 ; 2^n entries (n==PAWHNHASHSIZE)
; one entry needs 8 Byte
;17=1MB, 16=524K, 15=256K, 14=128K, 13=64K, 12=32K
; The size of the materialhashtable
MATERIALHASHSIZE = 15 ; 2^n entries (n==MATERIALHASHSIZE)
; one entry needs 8 Byte
;12=32K, 11=16K, 10=8K, 9=4K, 8=2k
; The size of the evaluationhashtable
; currently not used!
EVALHASHSIZE = 1 ; 2^n entries (n==EVALHASHSIZE)
; one entry needs 8 Byte
Best wishes
Volker

Thanks Volker,
I had forgotten by now that I had originally set Gromit up too quickly this summer when the new version showed up, I'll get a fresh download of Gromit and set it up again and run it's games over again. The commented sections in your eval.cfg file are clearer than the eval.cfg that came with Gromit, I'll save that to a text file, thanks.
Pete

[Volker Pittlik]

Geschrieben von:/Posted by: Volker Pittlik am 20. Dezember 1999 22:13:57:
Als Antwort auf:/As an answer to: Re: I'll have to rerun the Gromit games geschrieben von:/posted by: Pete Galati am 20. Dezember 1999 20:49:36:

Thanks Volker,
I had forgotten by now that I had originally set Gromit up too quickly this summer when the new version showed up, I'll get a fresh download of Gromit and set it up again and run it's games over again. The commented sections in your eval.cfg file are clearer than the eval.cfg that came with Gromit, I'll save that to a text file, thanks.
Pete

You're welcome. But it's nearly the original by Frank Schneider. I have only added the lines with calculations of the size of the tables. Without any warranty. But I think it's correct.
Volker

[Pete Galati]

Geschrieben von:/Posted by: Pete Galati am 20. Dezember 1999 22:20:21:
Als Antwort auf:/As an answer to: Re: I'll have to rerun the Gromit games geschrieben von:/posted by: Pete Galati am 20. Dezember 1999 20:49:36:
I'll run Gromit like this:
TRANSPOSITIONSIZE = 19 ; 2^n entries (n==TRANSPOSITIONSIZE)
; NEW: one entry now needs 16 byte (was 12 byte)
;22=64MB, 21=32MB, 20=16MB, 19=8MB, 18=4MB, 17=2MB, 16=1MB, 15=524K, 14=256K
; The size of the pawnhashtable is far less important than the
; size of the transposition-table. Values above 18 (2^18 entries)
; do not make much sense.
PAWNHASHSIZE = 17 ; 2^n entries (n==PAWHNHASHSIZE)
; one entry needs 8 Byte
;17=1MB, 16=524K, 15=256K, 14=128K, 13=64K, 12=32K
; The size of the materialhashtable
MATERIALHASHSIZE = 12 ; 2^n entries (n==MATERIALHASHSIZE)
; one entry needs 8 Byte
;12=32K, 11=16K, 10=8K, 9=4K, 8=2k
; The size of the evaluationhashtable
; currently not used!
EVALHASHSIZE = 19 ; 2^n entries (n==EVALHASHSIZE)
; one entry needs 8 Byte
All of the Gromit games will be removed and played over again, Gromit was going to play after Phalanx does in TSCP anyhow, it'll just have more games to play, this will take a while. Gromit will now have a little more hashtable than Crafty, but it's not paging the hardrive yet.
Pete

[Pete Galati]

Geschrieben von:/Posted by: Pete Galati am 21. Dezember 1999 00:01:41:
Als Antwort auf:/As an answer to: Re: I'll have to rerun the Gromit games geschrieben von:/posted by: Volker Pittlik am 20. Dezember 1999 22:13:57:

Thanks Volker,
I had forgotten by now that I had originally set Gromit up too quickly this summer when the new version showed up, I'll get a fresh download of Gromit and set it up again and run it's games over again. The commented sections in your eval.cfg file are clearer than the eval.cfg that came with Gromit, I'll save that to a text file, thanks.
Pete

You're welcome. But it's nearly the original by Frank Schneider. I have only added the lines with calculations of the size of the tables. Without any warranty. But I think it's correct.
Volker

Phalanx has finished playing, all Gromit games have been removed and will be played next with repaired cfg file.
Pete

Results from file knightx.pgn:
No. Name Win Draw Loss Unf. Score Games %
----------------------------------------------------------------------
1 Comet-wb +38 =6 -10 *0 41.0 54 75.9%
2 TCB +37 =7 -10 *0 40.5 54 75.0%
3 Crafty +34 =11 -9 *0 39.5 54 73.1%
4 Phalanx +37 =5 -12 *0 39.5 54 73.1%
5 Knightx +25 =6 -23 *0 28.0 54 51.9%
6 GNUchess +6 =3 -21 *0 7.5 30 25.0%
7 Dragon311 +4 =2 -24 *0 5.0 30 16.7%
8 GNUches5 +4 =1 -25 *0 4.5 30 15.0%
9 Fortress +3 =2 -25 *0 4.0 30 13.3%
10 TSCP +0 =1 -29 *0 0.5 30 1.7%
Total Games: 210
White Wins: 108 (51.4%)
Black Wins: 80 (38.1%)
Draws: 22 (10.5%)
Unfinished: 0 (0.0%)


http://members.xoom.com/avochess/knightx.txt