Nice Jabba 1.0 <> Prédateur 2.2 game

[JuLieN]

I just ported Prédateur to the iPhone (it's of course still private, as I just put together a simple text interface : I'll have to build a nice GUI before it can hit the App Store), which was the occasion to correct numerous bugs I had no idea were there before (Apple's Xcode is packed with really nice tools!). As a result, Prédateur 2.2 seems to be significantly stronger than its predecessors and now beats up Jabba, despite CCRL lists Jabba as 230 Elo points stronger than my older Prédateur 2.0.

Actually, all the games I try end up as fast victories for Prédateur, often in a spectacular way, like this very nice game I want to share with you, that has a ravishing finish (no opening book used). It's the desktop version of Prédateur, both engines running on an iMac in tournament cadence (40moves/2hours):

[Event "?"]
[Site "?"]
[Date "2011.06.22"]
[Round "?"]
[White "Jabba 1.0"]
[Black "Predateur v2.2.1"]
[Result "0-1"]
[ECO "A00"]

1. Nc3 d5 2. Nf3 d4 3. Ne4 f5 4. Nc5 e5 5. Nb3 e4 6. Nfxd4 c5 7. Nxc5 Bxc5 8.
Nb3 Bd6 9. Nd4 Bxh2 10. Nxf5 Bxf5 11. Rxh2 Nc6 12. c4 Nf6 13. a4 e3 14. dxe3
Nb4 15. e4 Qxd1+ 16. Kxd1 O-O-O+ 17. Bd2 Nxe4 18. Rc1 Rxd2+ 19. Ke1 Rhd8 20. e3
Rxf2 21. Be2 Rd2 22. Rxh7 Rfxe2+ 23. Kf1 Ng3+ 24. Kg1 Rxg2# 0-1

Position before Prédateur's 13th move :
[d]r2qk2r/pp4pp/2n2n2/5b2/P1P1p3/8/1P1PPPPR/R1BQKB2 b Qkq a3 0 13
I guess most engines will find this move easily, but will YOU?

[Richard Allbert]

Luckily my current version of Jabba finds e3

Code: Select all

 Jabba 1.3 > uci

id name Jabba 1.3
id author Richard Allbert

option name Hash type spin default 32 min 16 max 1024
option name Ponder type check default false
option name OwnBook type check default false
option name UCI_Chess960 type check default false
option name ClearHash type button
option name UseFutility1 type check default true
option name UseFutility2 type check default true
option name Fultility1Margin type spin default 50 min 0 max 500
option name Fultility2Margin type spin default 500 min 0 max 20000
option name SearchNullmove type check default true
option name SearchIID type check default true
option name HashTableCuts type check default true
uciok
position fen r2qk2r/pp4pp/2n2n2/5b2/P1P1p3/8/1P1PPPPR/R1BQKB2 b Qkq a3
go infinite
info depth 2 seldepth 7 score cp 151  time 3 nodes 460 pv  d8d6 g2g3
info depth 3 seldepth 5 score cp 193  time 7 nodes 1062 pv  d8d6 g2g3 e8g8
info depth 4 seldepth 11 score cp 166  time 18 nodes 4334 pv  d8d6 g2g3 e8c8 e2e
3
info depth 5 seldepth 14 score cp 189  time 46 nodes 17341 pv  d8d6 h2h1 c6e5 e2
e3 e8g8
info depth 6 seldepth 16 score cp 175  time 302 nodes 142643 pv  c6e5 d1b3 d8c7
b3g3 e8c8 e2e3
info depth 7 seldepth 16 score cp 206  time 620 nodes 316467 pv  c6d4 b2b3 f6g4
h2h3 e8g8 f2f3 g4e5
info depth 8 seldepth 21 score cp 227  time 1237 nodes 650057 pv  c6d4 a1a3 d8c7
 h2h1 c7c4 a3c3 c4d5 e2e3
info depth 9 seldepth 22 score cp 323  time 3768 nodes 2021611 pv  e4e3 h2h4 e3f
2 e1f2 f6g4 h4g4 f5g4 d1b3 d8e7 f2g1 e8c8
info depth 10 seldepth 25 score cp 370  time 9507 nodes 5171040 pv  e4e3 a1a3 e3
f2 e1f2 f6g4 f2g1 g4h2 a3e3 e8f8 g1h2 d8h4 h2g1 h4c4 d2d3
info depth 11 seldepth 25 score cp 362  time 11920 nodes 6544570 pv  e4e3 a1a3 e
3f2 e1f2 f6g4 f2g1 g4h2 g1h2 d8h4 h2g1 h4c4 e2e3 c4g4 d1g4
info depth 12 seldepth 29 score cp 373  time 21056 nodes 11634975 pv  e4e3 a1a3
e3f2 e1f2 f6g4 f2g1 g4h2 a3e3 e8f8 g1h2 d8h4 h2g1 h4c4 d2d3 c4d4 d1c2
stop

nulltry 1738419 nullcuts 1446498 54.5829%
futility 1 8128170 futility 2 1353571
hashcuts 403032
fh 144045 fhf 2325037 ordering  94.1661%
nodes 4927531 qnodes 9660373  66.2218%
qcuts 5145328
checkext = 958002 pawnseventh = 22 intopawnend = 0
pvssearch = 5062817 pvsresearch = 145287 2.86969%

bestmove e4e3
unknown command

Nice job with the iPhone port...

Richard

[JuLieN]

Luckily my current version of Jabba finds e3

Good! Any release date? ^^

I tried a few engines. Stockfish will play 0-0 (like Jabba 1.0), and all other engine will chose either e3 or Nd4... at the notable exception of Komodo, the only engine that choses to play Ng4!

Nice job with the iPhone port...

Richard

Thx Richard! That wasn't trivial, as my initial code was full of win32 api calls. Now it's really portable (I even compiled an AmigaOS version ). The first step was to convert it to MacOSX, what I did a few months ago, and then to reduce its footprint and get rid of the few carbon calls I had used with the MacOSX version (any carbon call make your App rejected by Apple in the AppStore, even if it works...), so it could run on the iPhone. Also, as Prédateur is still in Free Pascal, it implied to compile a special version of the compiler that could use iOS headers, get integrated to XCode and produce binaries signed by Apple's keychain.

[Richard Allbert]

Good! Any release date? ^^

No, not in the near future. 1.3 is at least 200elo stronger than 1.0, bit I've become a bit disheartened with it all recently. I started a massive testing program, only to gain nothing in strength because the Time Control was too short and Games too few. Ended up in a big mess with 40+test versions, and no improvements.

Which makes it all the more galling when "new"engines come out with 3000elo

Plus I started working for myself this year and don't have the time that I did...

However,
What is always nice is going to the OTB competitions like CPT, I'll keep doing that.

Good luck with the iPhone - I used XCode for the Mac verison of Jabba, it's a very nice IDE.

Richard

[Graham Banks]

Good! Any release date? ^^

No, not in the near future. 1.3 is at least 200elo stronger than 1.0, bit I've become a bit disheartened with it all recently. I started a massive testing program, only to gain nothing in strength because the Time Control was too short and Games too few. Ended up in a big mess with 40+test versions, and no improvements.

Which makes it all the more galling when "new"engines come out with 3000elo

Plus I started working for myself this year and don't have the time that I did...

However,
What is always nice is going to the OTB competitions like CPT, I'll keep doing that.

Good luck with the iPhone - I used XCode for the Mac verison of Jabba, it's a very nice IDE.

Richard

There will always be true enthusiasts looking forward to the release of a new Jabba, no matter how weak or strong it is.

Hopeful regards,
Graham.

[royb]

Luckily my current version of Jabba finds e3

Good! Any release date? ^^

I tried a few engines. Stockfish will play 0-0 (like Jabba 1.0), and all other engine will chose either e3 or Nd4... at the notable exception of Komodo, the only engine that choses to play Ng4!

Nice job with the iPhone port...

Richard

Thx Richard! That wasn't trivial, as my initial code was full of win32 api calls. Now it's really portable (I even compiled an AmigaOS version ). The first step was to convert it to MacOSX, what I did a few months ago, and then to reduce its footprint and get rid of the few carbon calls I had used with the MacOSX version (any carbon call make your App rejected by Apple in the AppStore, even if it works...), so it could run on the iPhone. Also, as Prédateur is still in Free Pascal, it implied to compile a special version of the compiler that could use iOS headers, get integrated to XCode and produce binaries signed by Apple's keychain.

Actually Stockfish-2.1.1 64-bit using one core plays 13. ... Ng4 also after 49 seconds on my setup (Mac OS X 2.3 GHz i7 running Ubuntu 11.04 as a virtual machine):

1 -2.90 13.... O-O (0.30)
2 -2.90 13.... O-O 14.b3 (0.30)
3 -2.98 13.... O-O 14.c5 Ng4 15.Qb3+ Rf7 (0.30)
4 -3.43 13.... Ng4 14.Rh1 O-O 15.Qb3 (0.30)
5 -3.63 13.... Ng4 14.Rh1 O-O 15.f3 exf3 16.gxf3 (0.30)
6 -3.83 13.... Ng4 14.Rh1 O-O 15.f3 Qd6 16.Rh3 exf3 17.gxf3 (0.30)
7 -3.83 13.... Ng4 14.Rh1 O-O 15.f3 Qd6 16.Rh3 exf3 17.gxf3 Nf6 (0.31)
8 -3.83 13.... O-O 14.Rh1 Be6 15.Qb3 Nd4 16.Qc3 Ng4 17.f3 (0.32)
9 -3.95 13.... O-O 14.f4 Ng4 15.Rh3 e3 16.dxe3 Qxd1+ 17.Kxd1 Nf2+ 18.Ke1 Nxh3 19.gxh3 (0.34)
10 -4.24 13.... O-O 14.Rh1 Be6 15.Qb3 Nd4 16.Qc3 Ng4 17.f3 exf3 18.gxf3 Qf6 19.fxg4 Bxg4 (0.38)
11 -5.09 13.... O-O 14.g3 Ng4 15.Rg2 Be6 16.d3 Nce5 17.Bf4 exd3 18.exd3 Qa5+ 19.Bd2 Qc5 (0.54)
12 -5.93 13.... O-O 14.Rh1 Ng4 15.f3 Qd6 16.Rh3 exf3 17.gxf3 Nge5 18.Rh5 Bg6 19.Rh3 Nxc4 (0.91)
13 -6.30 13.... O-O 14.g3 Ng4 15.Rh4 Nce5 16.Qb3 Qd4 17.e3 Nf3+ 18.Kd1 Nxf2+ 19.Kc2 Qd7 20.Qb5 Nxh4 21.Qxd7 Bxd7 22.gxh4 Rac8 23.b3 (1.20)
14 -6.30 13.... O-O 14.g3 Ng4 15.Rh4 Nce5 16.Qb3 Qd4 17.e3 Nf3+ 18.Kd1 Nxf2+ 19.Kc2 Qd7 20.Qb5 Nxh4 21.Qxd7 Bxd7 22.gxh4 Rac8 23.b3 (1.50)
15 -6.90 13.... O-O 14.Rh4 Nd4 15.g3 e3 16.Bg2 Nc2+ 17.Kf1 Nxa1 18.d3 exf2 19.Bf4 Qb6 20.Qxa1 Rae8 21.e4 Bg4 22.Qc1 h5 (3.49)
16 -6.90 13.... O-O 14.Rh4 Nd4 15.d3 exd3 16.Rxd4 Qxd4 17.Be3 Qxb2 18.Qc1 Qb4+ 19.Qd2 Qxc4 20.exd3 Qd5 21.Qa2 Rf7 22.Qxd5 Nxd5 23.Rc1 Nxe3 24.fxe3 (5.66)
17 -7.83 13.... O-O 14.Rh4 Nd4 15.d3 exd3 16.Rxd4 Qxd4 17.Be3 Qxb2 18.Qc1 Qb4+ 19.Qd2 Qxc4 20.exd3 Qh4 21.Rb1 Rf7 22.g3 Qxa4 23.Qd1 Qxd1+ 24.Kxd1 Ng4 25.Bd4 Rd8 26.Rb4 (14.69)
18 -7.87 13.... O-O 14.Rh1 Ng4 15.Ra3 e3 16.Rxe3 Nxf2 17.Kxf2 Bc2+ 18.Ke1 Bxd1 19.Kxd1 Qa5 20.g3 Rae8 21.Bg2 Rxe3 22.Bd5+ Kh8 23.dxe3 Qxa4+ 24.Kd2 (19.13)
19 -8.68 13.... Ng4 14.Rh3 O-O 15.Rg3 Nd4 16.Rxg4 Bxg4 17.d3 Qa5+ 18.Bd2 Qb6 19.Bc3 exd3 20.Bxd4 Qxd4 21.f3 Rad8 22.exd3 Qxb2 23.Be2 (49.57)
20 -8.56 13.... Ng4 14.d4 exd3 15.Rh3 O-O 16.Be3 Nd4 17.exd3 Nxf2 18.Kxf2 Bg4+ 19.Kg1 Bxd1 20.Rxd1 Qf6 21.Kh1 Rae8 22.Rb1 Rf7 23.Rd1 Qe5 24.b4 h6 (68.28)
21 -8.84 13.... Ng4 14.Rh1 O-O 15.Ra3 e3 16.fxe3 Nf2 17.Kxf2 Bc2+ 18.Ke1 Rxf1+ 19.Kxf1 Bxd1 20.Ke1 Bc2 21.Rc3 Nb4 22.g3 Qg5 23.e4 Bxe4 24.Rh4 Nd3+ (97.04)
22 -8.80 13.... Ng4 14.Rh3 O-O 15.d4 exd3 16.f3 Nd4 17.fxg4 Nc2+ 18.Kf2 Bxg4+ 19.Rf3 Bxf3 20.exf3 Qb6+ 21.Kg3 Nxa1 22.Qxd3 Rad8 23.Qe4 Nb3 24.Be3 Nd2 25.Bxb6 Nxe4+ 26.fxe4 axb6 27.Be2 Rd4 28.Bf3 Rxc4 29.b3 (155.70)

Nice to see Komodo coming up with this move also of course!

Roy