Winboard bullet tournament

[pete]

Geschrieben von: / Posted by: pete at 09 May 2000 22:34:55:
Hi,
at last inspired by this forum I thought I'd try a winboard tournament on my compi for myself .
As I mostly was interested in learning about possible inherent problems or mistakes in setup I chose a rather unusual time control : 150 minutes in 2 minutes .
Games played on a PIII500 with 128MB RAM , ponder off , all engines used best settings I know and about 16 MB for Hash ; the ones who are able to used 5-men tablebases .
And the winner was : Crafty ( what a surprise ) .
Interesting things happened and also I learnt a lot about winboard tournaments in general and some engines I never tried before . I will later comment on this tournament and my thoughts but for tonight I try to post the result and the crosstable ( as I am nearly unaware of HTML a major challenge ) .
It was a Swiss tournament ; 13 rounds à 4 games ( means 52 games for everyone ).
Only thing I better add _now_ : there were about a handful of programs who couldn't deal with this time control at all : main example amy ( which I know as a very strong program at more usual ones ) , chop , Francesca ( which was a disappointment as I chose the n moves in m minutes just for it as I didn't know it before ), sos ( partly ) and ant .
But when compaired to my private blitz results at longer time controls usually ( played manually on two computers with ponder on ) this list correlates surprisingly well .
Ok ; for now enough of my crappy English : let's try the HTML thing :
Standings
Place Name Feder Rtg Loc Score Buch. Progr. M-Buch.
1 Crafty17.10, 40 399.5 269.5 334.0
2 Shredder3 wb, 39 400.0 255.0 333.5
3 Nimzo2000b, 35.5 407.5 256.0 340.5
4 Comet_b21, 34 404.0 232.0 338.5
5 Anmon5.06, 33.5 386.5 227.5 335.0
6-7 LG20002.7, 30.5 387.5 208.0 336.0
tcb045, 30.5 381.0 206.5 340.0
8 Gromit3.0, 30 373.5 200.5 322.0
9-10 phalanx22, 29.5 404.5 227.0 338.0
bringer1.5, 29.5 344.0 197.0 304.0
11 Inmi3.03, 29 305.0 182.0 264.0
12-14 Exchess3.14, 27.5 355.0 203.5 308.5
Gnuchess4, 27.5 325.0 200.5 283.0
bringer1.6, 27.5 321.5 188.5 279.5
15-18 Gromit2.20, 27 353.5 184.5 302.0
Gnuchess5, 27 316.0 183.5 275.0
Arasan5.3, 27 292.5 190.5 258.0
Averno0.28, 27 289.0 176.5 248.0
19 SOS, 26.5 356.0 197.5 315.0
20 Knightx1.53, 25.5 271.0 167.0 236.0
21 Fortress1.61, 24 290.5 174.5 256.0
22 Lambchop7.1, 17 273.0 125.5 243.0
23 Zephyr, 12.5 278.0 102.0 248.0
24 Francesca, 11.5 296.5 105.0 262.0
25 Ant4.16, 6.5 284.0 64.0 254.0
26 Amy 0.7, 1 293.5 8.0 256.5


Cross Table
No Name Feder Rtg 1 2 3 4 5 6 7 8 9 10 11 12 13
1. Amy 0.7, 14:0.0 17:0.0 13:0.0 21:0.0 15:0.0 6:1.0 23:0.0 20:0.0 9:0.0 26:0.0 22:0.0 25:0.0 3:0.0
2. Gnuchess4, 15:4.0 14:4.0 7:2.0 11:0.0 5:1.5 19:1.5 18:1.0 26:1.5 6:4.0 25:2.0 20:1.0 9:1.0 13:4.0
3. Gnuchess5, 16:0.0 15:3.0 22:3.5 19:2.0 9:2.5 10:.5 12:1.5 13:4.0 11:1.0 6:4.0 4:1.0 18:0.0 1:4.0
4. Crafty17.10, 17:3.5 10:3.0 8:3.0 7:2.5 11:3.0 16:1.5 9:3.0 18:3.5 5:3.0 19:3.5 3:3.0 20:3.5 23:4.0
5. LG20002.7, 18:1.0 24:3.5 14:3.0 8:2.0 2:2.5 17:1.0 26:3.0 19:3.0 4:1.0 16:1.0 21:4.0 7:4.0 10:1.5
6. Zephyr, 19:0.0 21:2.0 23:0.0 15:1.5 13:3.5 1:3.0 24:0.0 22:0.0 2:0.0 3:0.0 14:2.5 12:0.0 20:0.0
7. Nimzo2000b, 20:4.0 16:3.0 2:2.0 4:1.5 10:3.0 11:4.0 8:1.5 17:3.0 12:3.0 18:4.0 25:3.0 5:0.0 24:3.5
8. Anmon5.06, 21:4.0 19:2.5 4:1.0 5:2.0 24:3.0 18:2.5 7:2.5 16:3.0 10:1.5 17:1.5 11:2.5 14:3.5 26:4.0
9. SOS, 22:3.5 26:2.0 19:2.0 18:1.5 3:1.5 24:3.5 4:1.0 11:0.0 1:4.0 14:4.0 16:0.0 2:3.0 17:.5
10. Comet_b21, 23:3.0 4:1.0 20:4.0 16:2.0 7:1.0 3:3.5 17:2.5 25:4.0 8:2.5 11:2.5 24:2.5 22:3.0 5:2.5
11. phalanx22, 24:2.5 12:3.5 26:4.0 2:4.0 4:1.0 7:0.0 16:.5 9:4.0 3:3.0 10:1.5 8:1.5 17:2.5 25:1.5
12. Gromit2.20, 25:1.0 11:.5 15:4.0 14:1.0 20:3.0 23:2.5 3:2.5 24:2.5 7:1.0 22:4.0 17:1.0 6:4.0 16:0.0
13. Ant4.16, 26:0.0 20:0.0 1:4.0 22:0.0 6:.5 21:1.5 15:.5 3:0.0 14:0.0 23:0.0 19:0.0 24:0.0 2:0.0
14. Fortress1.61, 1:4.0 2:0.0 5:1.0 12:3.0 22:2.0 26:1.5 20:2.0 23:1.5 13:4.0 9:0.0 6:1.5 8:.5 15:3.0
15. Lambchop7.1, 2:0.0 3:1.0 12:0.0 6:2.5 1:4.0 20:0.0 13:3.5 21:4.0 22:0.0 24:0.0 23:1.0 26:0.0 14:1.0
16. Shredder3 wb, 3:4.0 7:1.0 25:3.5 10:2.0 17:3.0 4:2.5 11:3.5 8:1.0 20:4.0 5:3.0 9:4.0 19:3.5 12:4.0
17. tcb045, 4:.5 1:4.0 18:2.0 23:4.0 16:1.0 5:3.0 10:1.5 7:1.0 24:3.0 8:2.5 12:3.0 11:1.5 9:3.5
18. Gromit3.0, 5:3.0 25:2.0 17:2.0 9:2.5 19:2.0 8:1.5 2:3.0 4:.5 23:4.0 7:0.0 26:1.5 3:4.0 21:4.0
19. Exchess3.14, 6:4.0 8:1.5 9:2.0 3:2.0 18:2.0 2:2.5 25:1.5 5:1.0 26:4.0 4:.5 13:4.0 16:.5 22:2.0
20. Inmi3.03, 7:0.0 13:4.0 10:0.0 25:2.5 12:1.0 15:4.0 14:2.0 1:4.0 16:0.0 21:4.0 2:3.0 4:.5 6:4.0
21. Francesca, 8:0.0 6:2.0 24:0.0 1:4.0 26:0.0 13:2.5 22:3.0 15:0.0 25:0.0 20:0.0 5:0.0 23:0.0 18:0.0
22. Knightx1.53, 9:.5 23:2.5 3:.5 13:4.0 14:2.0 25:0.0 21:1.0 6:4.0 15:4.0 12:0.0 1:4.0 10:1.0 19:2.0
23. Averno0.28, 10:1.0 22:1.5 6:4.0 17:0.0 25:1.5 12:1.5 1:4.0 14:2.5 18:0.0 13:4.0 15:3.0 21:4.0 4:0.0
24. bringer1.6, 11:1.5 5:.5 21:4.0 26:3.5 8:1.0 9:.5 6:4.0 12:1.5 17:1.0 15:4.0 10:1.5 13:4.0 7:.5
25. bringer1.5, 12:3.0 18:2.0 16:.5 20:1.5 23:2.5 22:4.0 19:2.5 10:0.0 21:4.0 2:2.0 7:1.0 1:4.0 11:2.5
26. Arasan5.3, 13:4.0 9:2.0 11:0.0 24:.5 21:4.0 14:2.5 5:1.0 2:2.5 19:0.0 1:4.0 18:2.5 15:4.0 8:0.0


greetings to the winboarders
pete

[Frank Quisinsky]

Geschrieben von: / Posted by: Frank Quisinsky at 10 May 2000 19:08:49:
Als Antwort auf: / As an answer to: Winboard bullet tournament geschrieben von: / posted by: pete at 09 May 2000 22:34:55:
Hi,
I hope you can send me the PGN files for my tournament page.
Very interesting results !
Regards
Frank

[pete]

Geschrieben von: / Posted by: pete at 10 May 2000 20:25:53:
Als Antwort auf: / As an answer to: Re: Winboard bullet tournament geschrieben von: / posted by: Frank Quisinsky at 10 May 2000 19:08:49:

Hi,
I hope you can send me the PGN files for my tournament page.
Very interesting results !
Regards
Frank

Hi ,
sure I can send the pgn-file but I would have to do some cosmetic changes to make it readable .
For example both Bringers are in as "Bringer" and gromit2.20 is "engine" ; I never renamed the default names when setting them up as I knew who was who .
For nearly all engines you can't conclude the version number out of the name of the executable ( this would better be changed too ) .
Also I would have to take a few games out which had to be replayed ( Problems with Fortress mainly ).
Then there a few games saved in the same pgn-database which don't belong to the tournament.
Also I would like to change the entries for "Event" and "Site" for all games .
Is there a tool which allows to do that in a simple way under windows ? An URL for a freeware program would be appreciated or an explanation how it can be done with something like Fritz.
If I had to do it with notepad I better wait until next week when I am back to work and do it with sed on a UNIX box .
greetings.
pete

[Frank Quisinsky]

Geschrieben von: / Posted by: Frank Quisinsky at 10 May 2000 20:46:28:
Als Antwort auf: / As an answer to: Re: Winboard bullet tournament geschrieben von: / posted by: pete at 10 May 2000 20:25:53:

Hi ,
sure I can send the pgn-file but I would have to do some cosmetic changes to make it readable .
For example both Bringers are in as "Bringer" and gromit2.20 is "engine" ; I never renamed the default names when setting them up as I knew who was who .
For nearly all engines you can't conclude the version number out of the name of the executable ( this would better be changed too ) .
Also I would have to take a few games out which had to be replayed ( Problems with Fortress mainly ).
Then there a few games saved in the same pgn-database which don't belong to the tournament.
Also I would like to change the entries for "Event" and "Site" for all games .
Is there a tool which allows to do that in a simple way under windows ? An URL for a freeware program would be appreciated or an explanation how it can be done with something like Fritz.
If I had to do it with notepad I better wait until next week when I am back to work and do it with sed on a UNIX box .
greetings.
pete

Hm, I have not problems with the PGN files. I edit all games for my tournament page with Chess-Base 7.
Convert PGN to CBH, now I edit in CBH and convert CBH to PGN

For me is important:
- number from the version
- hashtable size
- tablebases yes/no 4-pieces or 5-pieces !
- time control
- ponder on/off
When you will you can send my a picture from your person or your comments for your tourney, excel files, or elostat, doc, txt files, what you will. When I have time I make then the tournament HTML page.
Yes but you can write me the version number per eMail or in a txt file. It`s not a problem for me. You can sent furthermore 1000 pgn files. Werner Inmann has a very good tool (PGN-Sammler, in my download page). So I have in seconds 1000 PGN files in 1 pgn file !
Brice Boissel sent me 2000 pgn files
It is not many work to make 1 pgn file.

Which problems !
I delete Fortress in my FQ-2 tournament. I can not play black games with Fortress or more then 1 game with mg x !

This is a problem for me. I hope you can delete the games in your files.
I think Chess-Base light is good !
And PGN Sammler !
Frank

[pete]

Geschrieben von: / Posted by: pete at 12 May 2000 20:20:27:
Als Antwort auf: / As an answer to: Winboard bullet tournament geschrieben von: / posted by: pete at 09 May 2000 22:34:55:
The tournament was played for pure fun without any special purpose ; time control chosen was 150 moves in 2 minutes as I was mainly interested in seeing several engines play for the 1st time on my computer and wanted to learn which problems you have to expect in "Winboard tournaments" on a single computer with "Ponder=Off" .
It was a 13x4 rounds swiss tournament .
And it was very interesting : I learned a lot and for most engines it seems that their ranking was surprisingly near to the results reported by others at longer time controls .
In general the quality of the games was ok ; there were a few quite exciting games but for sure also a few bugs and blunders ; I watched the games live (like television )
The result shouldn't be overestimated for sure ; there are several engines who simply weren't prepared for this unusual time control ( s.b. in comments )
Games were played on a PentiumIII , 128 MB RAM ; those who were able to could use all 3-/4-/5-men tablebases ; hash was set to approximately 16 MB for every engine .
Games were played using /debug /mg 4 /reuse /reuse2 . There are a few engines who also start their own GUI under winboard or have to initialize something at start and so lose up to 5-15 seconds at the start of every game ; the handicap didn't seem to matter too much but for sure it is there at such a short time control (s.b).
A few very personal live impressions :
1.) Crafty
Crafty was at the tops during the whole tournament ; it didn't have any problem with the time control ; its major strength was a superb endgame where it won most of its games . A well-deserved winner . Crafty won all its 4game-matches except a 1,5-2,5 loss against Shredder.
2.) Shredder3 wb
Shredder was very close and could have been the winner too ; no major problems with time control ; lost a few seconds at the start of every game as it starts its own GUI too ; style similar to crafty . Decisive was probably that Shredder lost 1-3 against Anmon .
3.) Nimzo2000b
Nimzo was long time a major contender for 1st place ; clearly tactically the strongest engine at this time control , had a little bad luck in the end.
Nimzo lost 11 games on time !! ( although most were in lost positions )but until round 11 could keep crafty's and shredder's pace but after a 0-4 loss against Little Goliath it was all over .
I got the impression that this engine will probably win most blitz tournaments if the time control is a litte longer.
4.) Comet-B21
Comet plays an enterprising and positional sound chess even at this short time control ; started bad but had a nice finish to the 4th place .
5.) Anmon5.06
No visible problems with time control . Anmon can win against everybody but unfortunately also lose against many . Very risky chess , fun to watch ; probably even stronger at longer time controls .
2,5-1,5 against Nimzo , 3,0-1,0 against Shredder !
6.) Little Goliath
No visible problems with time control ; played IMHO a quite ugly chess . Not knowing too much about positional play it couldn't reach the depths necessary to create tactical havoc . Only really good match was the one against Nimzo.
7.) The Crazy Bishop
Where others more or less work at this time control tcb seems to love it . tcb played with the fritz4-book , impressive tactical shots ; at this time control tcb is still very close to the top.
8.) Gromit 3.0
Gromit3.0 is different than 2.2 ; it searched quite deep and played nice enterprising tactical games . As well as tcb played with the fritz4-book , lost a few seconds at the start of every game but had no visible problems with time control . A little unstable in performance though ; good games were followed by weak ones . Won 3,0-1,0 against Little Goliath!
9.) Phalanx
The unlucky hero of this tournament . Was at the very top during the 1st half of the tournament . Phalanx went for the enemy king every time possible and played some very beautiful short games ; always a fun to watch ; when the position becomes boring though phalanx also commited quite a few positional mistakes ; while phalanx plays dashing against the weaker engines against the absolut tops it lost too many . A 1,5-2,5 loss against Bringer1.5 in the final round brought it to a relatively bad end rank .
10.) Bringer1.5
Very good result . Bringer starts its own GUI ( loses up to 15 seconds at start of game ) ; scored well against the weaker ones and stole some points from the top players ; as usual bringer played well against phalanx
11.) InmiChess 3.03
Inmi finished surprisingly well here although it lost 8 games on time !! Inmi searches quite deep in a fast time . Major problem is its really buggy opening book ie playing with Black against the Ruy it nearly always plays 1.e4 e5 2.Nf3 Nc6 3. Bb5 Nge7 so in most games Inmi starts with a serious handicap . Also with white it often goes for things like a e4,d3,c4 - pawn structure . I suspect with a better book and better handling of time control it could have ended even higher ; lacks still understanding of pawn structures at times.
12.) Exchess3.14
This engine played very entertaining ; good active positional chess . Shoud have ended higher ( had a little bad luck sometimes ) ; weaker in endgame .
13.) GnuChess4
A good allrounder still ; able to steal points from many . Still a major challenge for the amateur programmers it seems . 2.0-2.0 against Nimzo !
14.) Bringer1.6
Comments similar to Bringer 1.5 ; a little slower than 1.5 I already expected it to have more difficuties at this pace .
15.) Gromit2.20
Well, well . My favourite Winboard program this is but it is simply too slow to play good bullet games . In a few games it showed its good positional understanding but commited too many tactical mistakes : won 3-1 against Inmi and 2,5-1,5 against Bringer1.6
16.) Gnuchess5.0
Tactically strong this program also showed several positionally unsound moves ; seems to need more tuning still . Got a 1,0-3,0 against crafty ! and won against sos 2,5-1,5 . Both GNUs seemed to handle this time control very well btw.
17.) Arasan 5.3
No problems with the time control ; won against the weak , lost against the good ones .
18.) Averno0.28
Averno played quite well at this time control ; solid and sensible moves ; no high search depths but the chess looked reasonable . I was quite surprised by Avernos good performance ; even had a little bad luck in some games .
19.) sos
With this rank starts the problem children of this tournament ; no visible problems with the time control were seen but sos really played bad ; this time control is simply too fast for this great program . Best results a 3,0-1,0 against Gnu4 and a 1,0-3,0 against crafty .
20.) Knightx
Lost 20 games on time!! Seems to be a good middleclass engine but couldn't deal with the time control .
21.) Fortress 1.61
Hmm ; I have to admit I love this program but it really always caused problems ; Fortress sometimes refuses to play as black against certain programs ; sometimes it is ok after restart , sometimes it simply refuses forever . When this problem happened I tried to restart 1 or 2 times and when Fortress then still wouldn't play it was a forfeit . In the end Fortress only played 45 of the 52 games . Also Fortress lost 6 games on time .
But when it was playing it was always fun ; I am looking forward to see it at longer time controls as it played some very beautiful king attacks and has a unique way of transforming every better position into an interesting pawn endgame . Probably stronger at longer time controls ?!
The time handling was extreme ( reminds me of one club member of my chess club ; played very slow ,but in the last 3 seconds was then able to get some 50 moves in , thrilling .
When it succeeds with its play it is usually beautiful : 1,0-3,0 LGG , won Gromit2.20 3,0-1,0
22.) Lambchop 7.1
Didn't use more than a few seconds for the whole game ; as a slow searcher then it didn't have any chance . Not designed for this time control !
23.) Zephyr
Positive surprise . Played quite good positional moves at times ; doesn't search very deep yet . Lost 10 games on time ! Often was able to hold itself very long even against the better programs .
24) Francesca
Francesca didn't work at this time control ; at least it didn't lose material in 1 ply ( s.b.) and not resigning it got a few points , played instantly . I was very disappointed as some engines would have done probably better at 2/0 and I made this strange 150/2 just for it but well : longer time controls will show Francesca stronger for sure ; so maybe another time .
25.) Ant4.16
Very similar to Francesca this engine had the additional problem that it resigned when down in material ; therefor less points out of luck .
26.) Amy0.7
Amy simply can't play at this time control ; it doesn't even look at the 1-ply material threats ; it seems to just take a legal move and do it . Got one point by winning on time against poor Zephyr.