Winboard Forum

I can not understand why some engines such as
Shredder 8 ,Gandalf 6.0...are playing differently under other GUI's when they
are exactly with same conditions ?

I wonder: under which GUI- the engines choose its best,strongest move ?

Important:On this PC( Intel Celeron 1.8 GHz)for the first time is playing chess
engines !!!

Here its available more examples with other engines where you can see two
different games :

Intel Celeron 1.8 GHz
Winboard 4.2.7 GUI
40 moves in 5 minutes
32 mb hashtable size
Nunn openning -C88
Ponder off

[Event "Winboard 4.2.7"]
[Site "Celeron 1.8 GHz"]
[Date "2005.01.07"]
[Round "1"]
[White "WildCat 4"]
[Black "Comet B.68"]
[Result "*"]
[TimeControl "40/300"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 d6 8.
d3 Na5 9. Nc3 Nxb3 10. axb3 Bg4 11. h3 Bh5 12. Re3 O-O 13. Ne2 Bxf3 14.
Rxf3 d5 15. exd5 Nxd5 16. Ng3 Qd7 17. Qe2 f6 18. Bd2 Bc5 19. Ne4 Bd4 20. c3
Ba7 21. Kh2 Qe6 22. b4 Rfd8 23. Re1 Ne7 24. Be3 Bxe3 25. fxe3 a5 26. bxa5
Rxa5 27. Nc5 Qc6 28. d4 Ra2 29. Rf2 Nd5 30. Qf3 Rda8 31. Nd3 exd4 32. Nf4
Ne7 33. Qxc6 Nxc6 34. exd4 b4 35. Kg3 R8a5 36. Re6 Rg5+ 37. Kf3 Ra6 38. h4
Rga5 39. Re3 Kf7 40. Rfe2 bxc3 41. bxc3 Rb6
*

------------------------------------------------------
Intel Celeron 1.8 GHz
Arena 1.1 GUI
40 moves in 5 minutes
32 mb hashtable size
Nunn openning -C88
Ponder off



[Event "Arena 1.1"]
[Site "Celeron 1.8 GHz"]
[Date "2005.01.07"]
[Round "1"]
[White "WildCat 4.0"]
[Black "Comet B68"]
[Result "1/2-1/2"]
[ECO "C88"]
[Opening "Spanish"]
[Time "15:58:22"]
[Variation "Closed 7...d6"]
[TimeControl "40/300:40/300:40/300"]
[Termination "normal"]
[PlyCount "81"]
[WhiteType "program"]
[BlackType "program"]

{C:\Documents and Settings\PC05\Belgelerim\Nunn_draw\book.pgn 1/1120
(Player-Player, SedatChess, 2003.11.25, ?-?)} 1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4
Nf6 5.O-O Be7
6.Re1 b5 7.Bb3 d6 8.d3 Na5 9.Nc3 Nxb3 10.axb3 Bg4 11.h3 {-0.20/10} Bh5 {-0.61/9}
12.Re3 {-0.20/11} O-O {-0.23/9} 13.Ne2 {-0.19/11} Bxf3 {-0.09/9} 14.Rxf3
{+0.06/10} d5
{-0.18/10} 15.exd5 {+0.05/11} Nxd5 {-0.22/10} 16.Ng3 {+0.10/10} Qd7 {-0.15/9}
17.Qe2
{+0.13/10} f6 {-0.20/8} 18.Bd2 {+0.16/9} Bc5 {-0.20/8} 19.Ne4 {+0.15/9} Bd4
{-0.22/8}
20.c3 {+0.33/10} Ba7 {-0.31/8} 21.Kh2 {+0.26/9} f5 {-0.29/8} 22.Ng3 {+0.32/10}
Qe6
{-0.31/8} 23.c4 {+0.27/10} Ne7 {-0.17/8} 24.Bb4 {+0.17/9} c5 {-0.06/8} 25.Bd2
{+0.06/9}
Ng6 {+0.09/8} 26.Nh5 {-0.03/9} Bb8 {+0.15/8} 27.Kg1 {+0.04/9} Bd6 {+0.09/7}
28.Re1
{+0.06/9} Qd7 {+0.24/7} 29.Re3 {-0.51/9} f4 {+0.68/9} 30.Re4 {-1.41/10} Nh4
{+1.49/8}
31.Qg4 {-0.89/9} Qxg4 {+1.81/9} 32.hxg4 {-0.90/9} g6 {+1.35/8} 33.Ng3 {-0.69/10}
bxc4
{+1.00/8} 34.dxc4 {-0.38/11} Rfb8 {+0.73/9} 35.Bc3 {-0.29/11} Rxb3 {+0.87/9}
36.Nf1
{+0.02/11} Rb7 {+0.58/8} 37.Nh2 {+0.47/10} Rf8 {+0.63/9} 38.Rf1 {+0.36/10} g5
{+0.83/8}
39.Rd1 {+0.10/12} Rb6 {+0.49/9} 40.Bxe5 {+0.41/12} Re8 {+0.52/10} 41.Rde1
{+0.32/12}
{?-? User Adjudication} 1/2-1/2

------------------------------------------------------



Best,
Sedat

Even if the GUI settigns and all engine settings are in fact indentical, I think 40/5 is pretty fast, and that timing differences could easily account for slightly different moves being picked at certain points during the game (finishing a fail high/low for instance).

Not sure this can be "fixed", other than perhaps by playing many games (200?) in a match and looking at average results.

There are engines which have some built-in random. Not necessarily in terms of small evaluation differences (possible), but also different time consumption then and when (for one move) in identical conditions, which may also lead to another move. For example, that has been said about Junior in the past, which presumably tries to avoid out-booking by that.

Before comparing GUIs like that, try to reproduce a game in the same GUI with identical conditions. I'm sure this isn't always possible either.

Brian Richardson wrote:Even if the GUI settigns and all engine settings are in fact indentical, I think 40/5 is pretty fast, and that timing differences could easily account for slightly different moves being picked at certain points during the game (finishing a fail high/low for instance).

Not sure this can be "fixed", other than perhaps by playing many games (200?) in a match and looking at average results.

----------------------------------------------------------------------------

Ok.Here are more examples -30 minute game per side

Maybe now i can proof something:

[Event "Arena 1.1"]
[Site "AMD Athlon 2400 @2.21GHz 1024 DDRAM"]
[Date "2005.01.08"]
[Round "1"]
[White "Thinker 4.7a"]
[Black "Delfi 4.5"]
[Result "1-0"]
[ECO "C07"]
[Opening "French"]
[Time "00:34:25"]
[Variation "Tarrasch, Open, 4.Ngf3 Nf6"]
[TimeControl "1800+0"]
[Termination "normal"]
[PlyCount "90"]
[WhiteType "program"]
[BlackType "program"]

{C:\chess\Nunn\test\C07.pgn 1/1 (Thinker 4.7a-Delfi 4.5, AMD Athlon 2400
@2.21GHz1024 , 2005.01.07, 1-0)} 1.d4 e6 2.e4 d5 3.Nd2 c5 4.Ngf3 Nf6 5.exd5 Qxd5
6.Bc4 Qd8
7.O-O cxd4 8.Nb3 Nc6 9.Nbxd4 Nxd4 10.Nxd4 a6 11.Bf4 b5 12.Be2 {+1.00/9} Bb7
{-0.47/12} 13.c4 {+0.70/10} bxc4 {-0.46/11} 14.Rc1 {+0.51/10} Rc8 {-0.46/11}
15.Bxc4
{+0.69/10} Be7 {-0.67/11} 16.Qa4+ {+0.92/10} Qd7 {-0.69/11} 17.Bb3 {+0.80/11}
Bc5
{-0.65/12} 18.Rfd1 {+1.11/11} O-O {-1.21/12} 19.Qxd7 {+1.33/11} Nxd7 {-1.39/15}
20.Nxe6
{+1.29/14} fxe6 {-1.49/15} 21.Bxe6+ {+1.29/13} Kh8 {-1.45/15} 22.Bxd7 {+1.25/14}
Bxf2+
{-1.48/15} 23.Kxf2 {+1.32/14} Rxf4+ {-1.49/14} 24.Kg3 {+1.30/11} Rcf8 {-1.46/14}
25.Rc7
{+1.29/13} Rb4 {-1.48/13} 26.Rd2 {+1.50/10} Rd8 {-1.49/12} 27.a3 {+1.65/10} Rb3+
{-1.38/12} 28.Kf4 {+1.57/12} Rf8+ {-1.43/12} 29.Ke5 {+1.60/11} h6 {-1.41/11}
30.g3
{+1.53/11} Rb8 {-1.51/12} 31.Kd4 {+1.53/11} h5 {-1.63/12} 32.Rf2 {+1.70/12} Rb6
{-1.60/12}
33.b4 {+2.05/12} Rd6+ {-1.63/12} 34.Kc5 {+1.94/12} Rd1 {-1.97/13} 35.Rf5
{+2.11/13}
g6 {-2.07/13} 36.Rf6 {+2.04/12} Rc1+ {-2.30/12} 37.Kd6 {+2.33/12} Rd1+
{-2.47/12}
38.Ke5 {+2.27/11} Rd5+ {-2.47/11} 39.Ke6 {+2.28/12} Rd3 {-2.45/12} 40.Ke5
{+2.32/12}
Rd5+ {0.00/16} 41.Kf4 {+2.34/12} Kg7 {-2.45/12} 42.Rb6 {+2.50/12} Rf8+
{-2.70/12}
43.Bf5+ {+2.61/12} Kh8 {-2.70/12} 44.Kg5 {+2.80/13} Bc8 {-2.96/13} 45.Kh6
{+3.33/14}
Bxf5 {-3.48/16}
----------------------------------------------------------------------------

[Event "Winboard 4.2.7"]
[Site "AMD Athlon 2400 @2.21GHz 1024 DDRAM"]
[Date "2005.01.07"]
[Round "1"]
[White "Thinker 4.7a"]
[Black "Delfi 4.5"]
[Result "1-0"]
[TimeControl "1800"]
[PairNo ""]

1. d4 e6 2. e4 d5 3. Nd2 c5 4. Ngf3 Nf6 5. exd5 Qxd5 6. Bc4 Qd8 7. O-O cxd4
8. Nb3 Nc6 9. Nbxd4 Nxd4 10. Nxd4 a6 11. Bf4 b5 12. Be2 Bb7 13. c4 bxc4 14.
Rc1 Rc8 15. Bxc4 Be7 16. Qa4+ Qd7 17. Bb3 Bc5 18. Rfd1 O-O 19. Qxd7 Nxd7
20. Nxe6 fxe6 21. Bxe6+ Kh8 22. Bxd7 Bxf2+ 23. Kxf2 Rxf4+ 24. Kg3 Rcf8 25.
Rc7 Rb4 26. Rd2 Rd8 27. a3 Rb3+ 28. Kf4 Rf8+ 29. Ke5 h6 30. g3 a5 31. Ba4
Re3+ 32. Kd4 Re4+ 33. Kc5 Ba8 34. Bd7 Kh7 35. g4 Re7 36. Kb6 Re5 37. Bf5+
Kh8 38. Rcd7 Bf3 39. Rd8 Rg8 40. R2d6 Be2 41. b3 Rb5+ 42. Kc6 Rb8 43. Rxb8
Rxb8 44. b4 axb4 45. axb4 Kg8 46. Rd4 Bb5+ 47. Kc7 Re8 48. Re4 Rf8 49. Re5
Ba4 50. b5 Rf7+ 51. Kd6 Rb7 52. Bd7 Kh7 53. Bc6 Rb8 54. Kc7 Rf8 55. Be4+
Kh8 56. b6 Rf7+ 57. Kd6 Rd7+ 58. Kc5 Rd8 59. b7 Rb8 60. Kb6 Be8 61. Kc7
Rxb7+ 62. Kxb7 Bf7 63. Kc6
{Black resigns} 1-0

---------------------------------------------------------------------------

[Event "Hiarcs 9 GUI"]
[Site "AMD Athlon 2400 @2.21GHz 1024 DDRAM"]
[Date "2005.01.08"]
[Round "1"]
[White "Thinker 4.7a Wb2Uci"]
[Black "Delfi 4.5 Wb2Uci"]
[Result "*"]
[ECO "C07"]
[Annotator "1.56;-1.41"]
[PlyCount "63"]

{W=11.7 ply; 376kN/s B=12.3 ply; 1.031kN/s} 1. d4 e6 2. e4 d5 3. Nd2 c5 4.
Ngf3 Nf6 5. exd5 Qxd5 6. Bc4 Qd8 7. O-O cxd4 8. Nb3 Nc6 9. Nbxd4 Nxd4 10. Nxd4
a6 11. Bf4 b5 12. Be2 Bb7 13. c4 bxc4 14. Rc1 Rc8 15. Bxc4 Be7 16. Qa4+ Qd7 17.
Bb3 Bc5 18. Rfd1 O-O 19. Qxd7 Nxd7 20. Nxe6 fxe6 21. Bxe6+ Kh8 22. Bxd7 Bxf2+
23. Kxf2 Rxf4+ 24. Kg3 Rcf8 25. Rc7 Rb4 26. Rd2 Rd8 27. a3 {Both last book move
} Rb3+ {[%eval -141,12] [%emt 0:01:37]} 28. Kf4 {
(Kf2) [%eval 156,11] [%emt 0:00:24]} Rf8+ {(h6) [%eval -138,11] [%emt 0:00:32]}
29. Ke5 {(Kg4) [%eval 160,11] [%emt 0:00:22]} h6 {
[%eval -141,11] [%emt 0:00:33]} 30. Kd4 {(Ba4) [%eval 154,12] [%emt 0:00:26]}
Rf4+ {(Rf7) [%eval -131,12] [%emt 0:00:26]} 31. Kc5 {
[%eval 154,12] [%emt 0:00:22]} h5 {(Rf7) [%eval -140,13] [%emt 0:01:09]} 32.
Be8 {(g3) adjud. [%eval 165,10] [%emt 0:00:22]}


Best,
Sedat

Mike Scheidl wrote:There are engines which have some built-in random. Not necessarily in terms of small evaluation differences (possible), but also different time consumption then and when (for one move) in identical conditions, which may also lead to another move. For example, that has been said about Junior in the past, which presumably tries to avoid out-booking by that.

Before comparing GUIs like that, try to reproduce a game in the same GUI with identical conditions. I'm sure this isn't always possible either.

Yes maybe you are right about that there are engines which in identical conditions can lead to another move !!

But anyway its a strange that the mostly engines under each GUI are playing differently !

Best,
Sedat

Try fixed depth games instead of using time controls.
That might help.

Thanks for your interesting...

Yes but i must use time control becouse my goal is to choose exactly under which GUI will be more stabil for the winboard engines !

Yes but i must use time control becouse my goal is to choose exactly under which GUI will be more stabil for the winboard engines !

You don't need to spend a lot of time to find the GUI,
which is the most stable one for WB engines

Guenther

Yes i know what do you mean: Winboard GUI
How i understand for better quality games is the best way :
*Winboard engines under Winboard GUI
*Uci engines under Fritz,Arena,Shredder Clasic,ChessPartner...
*And the rest with autoplayer

Best,
Sedat

Sedat Canbaz wrote:

Brian Richardson wrote:Even if the GUI settigns and all engine settings are in fact indentical, I think 40/5 is pretty fast, and that timing differences could easily account for slightly different moves being picked at certain points during the game (finishing a fail high/low for instance).

Not sure this can be "fixed", other than perhaps by playing many games (200?) in a match and looking at average results.

----------------------------------------------------------------------------

Ok.Here are more examples -30 minute game per side

Maybe now i can proof something:

This isn't new, but in any case one must ask whether you turned off learning in the examples you gave.

Albert

Always while i am running engine-matches :
------------------------------------------

-No surfing on the Net,no working with Word,not doing some Programming etc.

-Before to start each tournament under both GUI's the PC is restarted !

-For each GUI was setuped separate (different)folder of engine.exe

-No any program is insalled in the TaskBar !

-Learning Files off