Beta Testers needed

[David Weller]

I uploaded my latest chess engine to

http://free-chess.home.comcast.net/xpdnt.exe // windows executable

If you want/need to see the source,

http://free-chess.home.comcast.net/xpdnt_0415.zip

Tis a winboard engine

Basic PVS with Apiration window [+-30]
Null Move R=2

check, mate threat, passed pawn, and one reply extensions
IID

12x16 board

no book
no egt
no ponder
no adjustable hash
no pawn hash

will resign
will offer draw


Looking forward to some/any feedback

There is a mysterious divide-by-zero bug which pops up now and then,
still looking ...

[Dann Corbit]

What does xpdnt stand for?

[David Weller]

It is supposed to sound like the word: eXPeDieNT

Code: Select all

Expedient

adj.
Appropriate to a purpose.

Serving to promote one's interest: was merciful only when mercy was expedient.
Based on or marked by a concern for self-interest rather than principle; self-interested.

Obsolete. Speedy; expeditious.


Funny thing is, my intention was to follow the 'obsolete' meaning and make a 'fast' engine

'based on a concern for self interest' may be more appropiate

or

'appropriate to the purpose' of driving me mad 8P

[Guenther Simon]

Hi David,

Thanks for your new program! I will test it in the next days.
BTW is GES' last version definitely the last one?

Oh another question , is XPDNT already able to play games from
fixed openings(PGN file) under Winboard('setboard' IIRC)?

Best regards,
Guenther

[David Weller]

Hi Guenther,

Yes, Xpdnt understands setboard.

I do hope to make one more revision of GES based on some things I learned writing Xpdnt - but who knows when that will happen

-David

[JohnBoyd]

I did some testing of Xpdnt on a P4 2.4ghz, 1-1 blitz, and it did pretty well, measuring about a 2125 rating after 2 small tournaments against a range of other engines in the Arena environment.

The nps was around 190k. It seems to be be able to move very quickly, often showing 40s left when the other engines had 15-20 seconds left.

It didn't crash once for me in either tourney.

What other specific tests would you like run?

thanks for the new contribution!

John

[David Weller]

Thanks John,

Just trying to catch any bugs and/or 'glaring' weaknesses

So your tourney helps!


-David

[Olivier Deville]

Hi David

Thanks a lot for this new engine It qualified easily for my next F tournament.

I have played a few games overnight. How to send them to you ?

Olivier

[David Weller]

Hi Olivier,

I updated my profile to show email

David

[David Weller]

I sent you my email address ...

[George Lyapko]

Which of your engines is stronger:GES or XPDNT?

It seems that XPDNT is better at least in endgames.Here are my results on modified MES suite(Athlon XP 1700+,20sec/pos):

55(48).xpdnt_0415-889(68 112 149 56 72 97 131 80 124)
85(75).GES_136-860(65 112 141 56 70 90 130 77 119)

[David Weller]

Hi George,

I dont know! Some time back [and many changes ago] GES was stronger. But not sure anymore.

[Guenther Simon]

Hello David,

I started a test for my GS15 matches, but I had to cancel it due
to time problems.
In both games Xpdnt lost on time, once in move 80 and once
in move 40. Tc was 40/15 on my P4 2.66Ghz under WB 4.27x.

Moreover I noticed two other problems. I counted 58! draw offers
in just two games, that is a way too much
(Thanks god the other engine didn't understand draw offers at all.
Normally I disable draw offer mechanisms anyway if adjustable
for several reasons)
BTW it made the draw offers practically at random scores,
even when it showed +5 or -100.
The next problem was that it a few times screwed up its score
randomly from +X to a sudden -100!(= mate against itself?)

Well at least it should have made 2/2 vs. Muse without the
timing glitch It seems already quite strong!

Here are the games(I will send the debug via mail, if I find
an address)

[Event "GS15 Test 19"]
[Site "ESPRESSO"]
[Date "2006.04.27"]
[Round "1"]
[White "Xpdnt_beta"]
[Black "Muse_0899b"]
[Result "0-1"]
[TimeControl "40/900"]
[Annotator "7. +0.27 7... +0.02"]

1. Nf3 d5 2. c4 e6 3. b3 Be7 4. Bb2 Bf6 5. Nc3 c5 6. cxd5 exd5 7.
e3 {+0.27/9} Ne7 {+0.02/10} 8. d4 {+0.31/9} cxd4 {-0.06/10} 9.
Nxd4 {+0.28/9} Nbc6 {-0.05/10} 10. Nxc6 {+0.19/10} Nxc6 {-0.04/10} 11.
Bd3 {+0.24/9} Qa5 {-0.10/10} 12. Rc1 {+0.49/9} d4 {-0.06/10} 13.
exd4 {+0.48/9} O-O {-0.22/10} 14. d5 {+0.49/9} Re8+ {+0.28/9} 15.
Kf1 {+0.77/10} Nb4 {+0.18/9} 16. a3 {+0.77/9} Nxd5 {-0.03/9} 17.
Bxh7+ {+0.45/9} Kxh7 {-0.18/9} 18. Nxd5 {+1.34/9} Qa6+ {-0.51/9} 19.
Rc4 {+1.24/9} Bxb2 {-0.51/9} 20. Nc7 {+1.62/9} Qg6 {-0.70/10} 21.
Nxe8 {+1.52/10} Bf5 {-0.26/10} 22. Rh4+ {+1.18/10} Kg8 {-0.92/6} 23.
Nc7 {-0.34/10} Bc2 {-0.35/9} 24. Nd5 {+0.68/9} Bxa3 {+0.65/9} 25.
Qf3 {+0.07/9} Re8 {+0.71/8} 26. g3 {-0.01/9} Be4 {+0.59/10} 27.
Rxe4 {-0.06/11} Qxe4 {+0.53/11} 28. Kg2 {-0.02/11} Qc2 {+0.47/10} 29.
b4 {+0.03/9} Bc1 {+0.49/10} 30. b5 {+0.20/9} Bb2 {+0.53/9} 31.
Rd1 {+0.23/9} Qc8 {+0.43/9} 32. Rd2 {+0.61/10} Bc1 {+0.00/10} 33.
Ra2 {+0.70/10} Qc5 {-0.10/10} 34. Qf5 {+0.89/9} Ba3 {-0.53/10} 35.
Qd7 {+1.07/9} Re1 {-0.92/10} 36. Ne3 {+1.47/9} Qf8 {-0.76/10} 37.
Qxb7 {+1.62/10} Bc5 {-0.83/10} 38. Rc2 {+1.55/10} Rb1 {-0.91/10} 39.
Qd7 {+1.69/10} g6 {-1.65/9} 40. Ng4 {+1.67/9} Re1 {-1.54/8} 41.
Nh6+ {+2.30/9} Kg7 {-2.26/11} 42. Nxf7 {+2.12/10} Qe7 {-2.25/9} 43.
Qd5 {+2.42/10} Bb6 {-2.30/10} 44. Nd6 {+2.41/10} Re5 {-2.39/9} 45.
Qd3 {+2.37/10} Bc5 {-2.39/10} 46. Ne4 {+2.33/10} Bb6 {-2.58/10} 47.
f3 {+2.58/9} Qb4 {-2.57/8} 48. Rc6 {+2.47/9} Qb2+ {-2.63/9} 49.
Nd2 {+2.62/10} Rf5 {-2.96/10} 50. Qd7+ {+2.75/10} Rf7 {-2.61/6} 51.
Qd6 {+2.58/10} Rf6 {-3.02/10} 52. Qd3 {+2.24/10} Ba5 {-2.57/10} 53.
Rc2 {+2.25/10} Qe5 {-2.58/10} 54. Ne4 {+2.85/9} Rf8 {-2.74/9} 55.
f4 {+2.98/9} Rd8 {-2.82/9} 56. fxe5 {+2.70/11} Rxd3 {-2.98/6} 57.
Nd6 {+2.76/10} Rd5 {-3.07/10} 58. Re2 {+2.67/12} Bb4 {-3.18/12} 59.
Ne8+ {+2.42/12} Kf7 {-2.86/12} 60. Nc7 {+2.46/12} Rd8 {-2.96/11} 61.
Kf3 {+2.40/11} Ba5 {-2.73/11} 62. Ra2 {+2.34/12} Bb6 {-2.70/11} 63.
Rc2 {+2.32/11} Rd3+ {-2.70/10} 64. Kf4 {+2.31/11} Kg7 {-2.87/10} 65.
Rc6 {+2.78/10} Rd2 {-2.98/9} 66. Kg5 {+2.94/10} Be3+ {-3.02/10} 67.
Kg4 {+2.96/10} Kh6 {-3.16/10} 68. h4 {+3.28/11} Rd4+ {-3.49/10} 69.
Kf3 {+3.53/12} Bd2 {-4.31/10} 70. Rd6 {+3.80/11} Bc3 {-4.29/11} 71.
g4 {+3.96/11} Bb2 {-4.59/10} 72. h5 {+4.97/11} Rxd6 {-4.67/11} 73.
exd6 {+4.34/12} Be5 {-5.33/14} 74. Ne8 {-100.01/13} Bd4 {-5.75/16} 75.
d7 {+5.38/11} Bb6 {-6.13/17} 76. Nd6 {+5.38/12} gxh5 {-6.97/14} 77.
gxh5 {+5.69/13} Kxh5 {-7.33/17} 78. Nf7 {-100.01/11} Kg6 {-6.91/14} 79.
d8=Q {+6.25/12} Bxd8 {-5.01/6}
{Black wins on time} 0-1

[Event "GS15 Test 19"]
[Site "ESPRESSO"]
[Date "2006.04.27"]
[Round "2"]
[White "Muse_0899b"]
[Black "Xpdnt_beta"]
[Result "1-0"]
[TimeControl "40/900"]
[Annotator "7. +0.09 7... +0.02"]

1. Nf3 d5 2. c4 e6 3. b3 Be7 4. Bb2 Bf6 5. Nc3 c5 6. cxd5 exd5 7.
Qc1 {+0.09/9} Kf8 {+0.02/9} 8. Ba3 {+0.64/9} Na6 {+0.23/9} 9. e3 {+0.61/10}
Nb4 {+0.39/10} 10. Be2 {+0.35/9} Bf5 {+0.49/10} 11. O-O {+0.32/10}
Ne7 {+0.46/10} 12. Qb2 {+0.24/10} Rc8 {+0.73/9} 13. d4 {+0.28/10}
Nc2 {+0.64/10} 14. Bxc5 {+0.20/10} Nxa1 {+0.56/10} 15. Qxa1 {+0.20/9}
h5 {+0.79/9} 16. Rc1 {+0.18/9} b6 {+0.47/10} 17. Bxe7+ {+0.11/10}
Bxe7 {+0.60/12} 18. Ba6 {+0.11/10} Rc7 {+0.38/12} 19. Nb5 {+0.21/11}
Rxc1+ {+0.53/11} 20. Qxc1 {+0.14/6} Rh6 {+0.39/11} 21. Ne5 {+0.22/11}
Bd6 {+0.20/10} 22. Nc6 {+0.65/11} Qa8 {+0.00/11} 23. Nxd6 {+0.64/11}
Rxd6 {-0.06/12} 24. Bb5 {+0.45/13} Rf6 {-0.13/12} 25. Qa3+ {+0.46/11}
Kg8 {-0.13/11} 26. Nxa7 {+0.53/11} Qb7 {-0.01/11} 27. b4 {+0.33/10}
h4 {+0.30/10} 28. f3 {-0.12/11} Re6 {+0.69/10} 29. Bf1 {-0.31/11}
Qe7 {+0.63/10} 30. Kf2 {-0.26/11} Qc7 {+0.65/10} 31. Nb5 {-0.55/11}
Qc2+ {+0.90/10} 32. Be2 {-0.83/11} Rc6 {+0.78/10} 33. Qa8+ {-1.04/10}
Rc8 {+0.95/11} 34. Qb7 {-0.76/11} Bd3 {+1.29/11} 35. Qxc8+ {-0.60/6}
Qxc8 {+1.67/14} 36. Bxd3 {-0.58/6} h3 {+1.69/13} 37. a3 {-1.54/14}
hxg2 {+1.99/13} 38. Kxg2 {-1.38/6} g5 {+2.07/13} 39. Be2 {-1.50/14}
Kg7 {+2.02/13} 40. h3 {-1.65/14}
{White wins on time} 1-0

Best regards,
Guenther

[David Weller]

Hi Guenther,

Thank you for you tests!

I have fixed the errors [I hope]

http://free-chess.home.comcast.net

[David Weller]

Thanks george,

Here is an update

http://free-chess.home.comcast.net

[David Weller]

Hello Olivier!

Thanks,

Here is an update!

http://free-chess.home.comcast.net

[Guenther Simon]

Hi Guenther,

Thank you for you tests!

I have fixed the errors [I hope]

http://free-chess.home.comcast.net

Fine! -> throwing new Xpdnt right into the test bed again

Best wishes,
Guenther

[Guenther Simon]

David Weller wrote:
Hi Guenther,

Thank you for you tests!

I have fixed the errors [I hope]

http://free-chess.home.comcast.net

Fine! -> throwing new Xpdnt right into the test bed again

Best wishes,
Guenther

Interesting it seems Xpdnt searches 1- 1.5 ply deeper now?
Is this only due to the bug fixes?

Best regards,
Gunther

[David Weller]

Interesting it seems Xpdnt searches 1- 1.5 ply deeper now?
Is this only due to the bug fixes?

I turned on 'late move reduction' [a kind of pruning] which is quite effective, because I use only null move pruning[R=2] otherwise

Most programs combine many different types of pruning and reducing methods

So basically, X is excluding many more lines from search now, enabling it to look a little deeper at them all

The jury is still out, as to whether this is a good thing ...

-David

[jshriver]

I'mt running a 5 0 match against crafty 19.15 on a Celeron 1.7 with 512meg ram on a linux box. And the result so far is Crafty 16, Xpdnt 2, Tie 1.

So multiple things to note: if it can beat crafty at all I find that wonderful for a beginning engine, and that it runs perfectly under linux using wine

xboard -size small -mg 25 -sgf xpdnt.pgn -fcp crafty -scp "wine ./xpdnt.exe"

I'll host the pgn when this is done if you want.

-Josh