Fruit 051103 vs. Toga II 1.3X4 41,5-18,5 (!!)

[Mike S.]

5m+2s, AthlonXP@1200 MHz
96 MB hash each, ponder off
Noomen 2006 opening set
Toga without bitbases
3-/4- and R vs. R-5-piece tbs. only

(I'm using the shorter name Fruit 2f for Fruit 05/11/03)

Code: Select all

Fruitoga 2007
                   
1  Fruit 2f       1011½½½111½½0½½1½½½11110½1½1½01½1½½½111½1½1111½½1½10½11½10½1  41.5/60
2  Toga II 1.3x4  0100½½½000½½1½½0½½½00001½0½0½10½0½½½000½0½0000½½0½01½00½01½0  18.5/60

29:6 wins and 25 draws. Performance difference 69,17% or 142 Elo.

I am aware that one single opponent, even less if it is a similar one, does not give a reliable result, BUT: Even if it's only half as much (~70 Elo) it would mean a huge improvement compared to 2.2.1.

A great result!

Example game, a quick win with black:

[Event "Fruitoga 2007"]
[Site "Schrotty"]
[Date "2007.07.24"]
[Round "10"]
[White "Toga II 1.3x4"]
[Black "Fruit 2f"]
[Result "0-1"]
[ECO "C88"]
[Annotator "0.15;0.16"]
[PlyCount "66"]
[TimeControl "300+2"]

{W=12.6 ply; 366kN/s B=13.4 ply; 364kN/s} 1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4
Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O 8. h3 Bb7 9. d3 d6 10. a3 Na5 11. Ba2 c5
12. Nbd2 Nc6 13. Nf1 Bc8 {Beide letzter Buchzug} 14. Ne3 {0.15/12 10} Be6 {
0.16/12 9} 15. Bd2 {0.17/11 8} Qd7 {0.24/12 14} 16. Nd5 {0.25/11 12} Qb7 {
(Tab8) 0.35/11 11} 17. Bg5 {0.27/10 9} Bxd5 {(Se8) 0.11/12 15} 18. exd5 {
0.46/12 12} Nd4 {0.01/13 11} 19. Nxd4 {0.50/13 11} cxd4 {0.13/12 8} 20. f4 {
(Lb3) 0.50/13 7} h6 {0.09/13 7} 21. fxe5 {0.31/13 8} hxg5 {0.06/14 12} 22. exf6
{0.29/12 6} Bxf6 {0.00/14 12} 23. Bb3 {0.28/13 12} Rae8 {(Le5) 0.00/15 7} 24.
a4 {0.26/13 9} Re3 {0.00/16 12} 25. axb5 {0.10/13 11} axb5 {-0.01/15 11} 26.
Qg4 {(Tf1) 0.00/14 10} Qe7 {-0.30/14 9} 27. Rf1 {-0.33/13 8} Be5 {-0.30/14 11}
28. Qd1 {(Tf2) -0.39/12 9} g6 {(Lg3) -0.65/12 6} 29. Ra5 {(La2) -0.74/12 9} f5
{(Lf4) -1.94/11 5} 30. Qd2 {-2.06/10 7} g4 {-2.84/12 10} 31. Rxb5 {
(Taa1) -3.06/10 17} gxh3 {-5.80/11 10} 32. g3 {-7.53/13 77} Bxg3 {-7.20/12 9}
33. Kh1 {-7.53/11 10} Re8 {(Te2) -12.20/12 10} 0-1

[Kaj Soderberg]

5m+2s, AthlonXP@1200 MHz
96 MB hash each, ponder off
Noomen 2006 opening set
Toga without bitbases
3-/4- and R vs. R-5-piece tbs. only

(I'm using the shorter name Fruit 2f for Fruit 05/11/03)

Code: Select all

Fruitoga 2007
                   
1  Fruit 2f       1011½½½111½½0½½1½½½11110½1½1½01½1½½½111½1½1111½½1½10½11½10½1  41.5/60
2  Toga II 1.3x4  0100½½½000½½1½½0½½½00001½0½0½10½0½½½000½0½0000½½0½01½00½01½0  18.5/60

29:6 wins and 25 draws. Performance difference 69,17% or 142 Elo.

I am aware that one single opponent, even less if it is a similar one, does not give a reliable result, BUT: Even if it's only half as much (~70 Elo) it would mean a huge improvement compared to 2.2.1.

Extraordinary !

I just started testing version 2.2.1 today, which looks very polished indeed, but seeing your result (OK, only blitz and only 1 opponent) i will very soon switch to testing the new potential monster.

Thanks for your quick effort.

Best regards,

Kaj

[Shaun]

Once some matches finish I will start testing this Fruit version for CCRL - if it is that much stronger and I hope it is - why was it not released earlier...

Shaun

[Eelco de Groot]

5m+2s, AthlonXP@1200 MHz
96 MB hash each, ponder off
Noomen 2006 opening set
Toga without bitbases
3-/4- and R vs. R-5-piece tbs. only

(I'm using the shorter name Fruit 2f for Fruit 05/11/03)

Code: Select all

Fruitoga 2007
                   
1  Fruit 2f       1011½½½111½½0½½1½½½11110½1½1½01½1½½½111½1½1111½½1½10½11½10½1  41.5/60
2  Toga II 1.3x4  0100½½½000½½1½½0½½½00001½0½0½10½0½½½000½0½0000½½0½01½00½01½0  18.5/60

29:6 wins and 25 draws. Performance difference 69,17% or 142 Elo.

I am aware that one single opponent, even less if it is a similar one, does not give a reliable result, BUT: Even if it's only half as much (~70 Elo) it would mean a huge improvement compared to 2.2.1.

A great result!

Example game, a quick win with black:

[Event "Fruitoga 2007"]
[Site "Schrotty"]
[Date "2007.07.24"]
[Round "10"]
[White "Toga II 1.3x4"]
[Black "Fruit 2f"]
[Result "0-1"]
[ECO "C88"]
[Annotator "0.15;0.16"]
[PlyCount "66"]
[TimeControl "300+2"]

{W=12.6 ply; 366kN/s B=13.4 ply; 364kN/s} 1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4
Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O 8. h3 Bb7 9. d3 d6 10. a3 Na5 11. Ba2 c5
12. Nbd2 Nc6 13. Nf1 Bc8 {Beide letzter Buchzug} 14. Ne3 {0.15/12 10} Be6 {
0.16/12 9} 15. Bd2 {0.17/11 8} Qd7 {0.24/12 14} 16. Nd5 {0.25/11 12} Qb7 {
(Tab8) 0.35/11 11} 17. Bg5 {0.27/10 9} Bxd5 {(Se8) 0.11/12 15} 18. exd5 {
0.46/12 12} Nd4 {0.01/13 11} 19. Nxd4 {0.50/13 11} cxd4 {0.13/12 8} 20. f4 {
(Lb3) 0.50/13 7} h6 {0.09/13 7} 21. fxe5 {0.31/13 8} hxg5 {0.06/14 12} 22. exf6
{0.29/12 6} Bxf6 {0.00/14 12} 23. Bb3 {0.28/13 12} Rae8 {(Le5) 0.00/15 7} 24.
a4 {0.26/13 9} Re3 {0.00/16 12} 25. axb5 {0.10/13 11} axb5 {-0.01/15 11} 26.
Qg4 {(Tf1) 0.00/14 10} Qe7 {-0.30/14 9} 27. Rf1 {-0.33/13 8} Be5 {-0.30/14 11}
28. Qd1 {(Tf2) -0.39/12 9} g6 {(Lg3) -0.65/12 6} 29. Ra5 {(La2) -0.74/12 9} f5
{(Lf4) -1.94/11 5} 30. Qd2 {-2.06/10 7} g4 {-2.84/12 10} 31. Rxb5 {
(Taa1) -3.06/10 17} gxh3 {-5.80/11 10} 32. g3 {-7.53/13 77} Bxg3 {-7.20/12 9}
33. Kh1 {-7.53/11 10} Re8 {(Te2) -12.20/12 10} 0-1

Incredibly strong result for Fruit 051103 Mike! I have trouble believing that Fruit 051103 could be much stronger than for instance Hiarcs 11 or Hiarcs 11.1 though... What could be happening is that in a match between two similar engines, Toga 1.3x4 is simply outsearched tactically because Fruit searches similar variations, only deeper... And Toga 1.3x4 is tactically probably just a bit less than Toga 1.2.1a as Marc Moisan stated, so Fruit may have the upper hand in tactics. But I've only played a few games with this version and mainly with some changed settings so I'm not sure exactly how strong 051103 really is...

Regards, Eelco

Some of the settings I did try for Fruit 051103 against Hiarcs 11:

The Gambit 001 settings for Fruit 051103 are very simple: everything stays the same but King Safety is increased from 100 to 125 as in the Toga Blueberry settings. I have tried more complicated changes but this one seems alright. If the defaults are much stronger though I retract everything and I will eat up my imaginary hat

Eelco

[ozziejoe]

I am not getting such good results thus far. Using the strategic opening test set, and 40 move in 4 minutes, repeating, toga 1.3x4 is winning thus far with 7 wins 3 losses and 15 ties. Still a ways to go though.