Thinker 5.2E blitz strength at S11 level (or above)

[Mike S.]

1m+1s, D945 3.4 GHz
128 MB hash tables
single engines, ponder off
Gambit Suite 2008 by Jeroen Noomen
Fritz 11 GUI, Vista 32

Code: Select all

1   Thinker 5.2e-32 passive  +38/-29/=33 54.50%   54.5/100 --> +33 Elo
2   Shredder 11 SE(*)        +29/-38/=33 45.50%   45.5/100

An excellent result! Thinker had a great start with 7.0/9. I was using Wb2Uci, but did not notice any technical problems this time (except the missing eval+depth infos). Thinker doesn't ponder though, in Fritz with Wb2Uci. I know the old Wb. adapter but I noticed, Thinker starts to ponder but stops during ponder time. So I do not trust it, and preferred a ponder off match.

The result is a mix of the performance from the gambit perspective and from the opponent's perspective; each variation was played twice. Due to style elements, it is of course possible that Thinker scores extra good versus Shredder in these gambit type positions, and maybe Shredder would score better against Thinker in more calm openings. Although, Shredder has been described as an aggressive engine too, so I do not think that this makes a big difference.

Despite the very short time control, the conditions seemed ok to me: Shredder achieved typical depths of 10...12 plies during the middlegame.

*) The S11 SE version does not use tablebases nor Shredderbases in the search. CEGT has this version in the 40/4 list. 33 Elo higher we find engines like Rybka 1.1 and 1.2f 32-bit or Shredder 11 64-bit. They have it also in the 40/20 list.

[Event "Thinkertest (single)"]
[Site "COMPUTER-11"]
[Date "2008.08.09"]
[Round "32"]
[White "Shredder 11 SE"]
[Black "Thinker 5.2e-32 passive"]
[Result "0-1"]
[ECO "C27"]
[PlyCount "44"]
[TimeControl "60+1"]

1. e4 e5 2. Nc3 Nf6 3. Bc4 Nxe4 4. Qh5 Nd6 5. Bb3 Nc6 6. Nb5 g6 7. Qf3 f5 8.
Qd5 Qe7 9. Nxc7+ Kd8 10. Nxa8 b6 {Both last book move} 11. Nxb6 {2.29/11 7}
axb6 {0.01/1 1} 12. Ne2 {2.01/11 2} Bb7 {0.01/1 2} 13. Qf3 {2.05/12 4} Nd4 {
0.01/1 3} 14. Qh3 {2.16/12 6} f4 {(Nxe2) 0.01/1 3} 15. Nxd4 {2.19/10 1} exd4+ {
0.01/1 3} 16. Kf1 {-1.79/11 9} Bg7 {(Bh6) 0.01/1 3} 17. Qg4 {-1.51/11 3} Re8 {
(f3) 0.01/1 3} 18. Be6 {-4.33/11 12} Nf5 {0.01/1 3} 19. Qd1 {-4.33/11 2} Qxe6 {
0.01/1 3} 20. f3 {-4.43/11 2} d3 {(Ba6+) 0.01/1 3} 21. c3 {-9.48/10 4} Bf6 {
(Nh4) 0.01/1 3} 22. b3 {-9.86/11 7} Nd4 {0.01/1 3} 0-1

[Event "Thinkertest (single)"]
[Site "COMPUTER-11"]
[Date "2008.08.09"]
[Round "80"]
[White "Shredder 11 SE"]
[Black "Thinker 5.2e-32 passive"]
[Result "0-1"]
[ECO "D43"]
[PlyCount "50"]
[TimeControl "60+1"]

1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. Nc3 e6 5. Bg5 h6 6. Bh4 dxc4 7. e4 g5 8. Bg3 b5
9. Be2 Bb7 10. h4 g4 11. Ne5 {Both last book move} b4 {0.01/1 2} 12. Na4 {
-0.43/12 5} Nxe4 {0.01/1 3} 13. O-O {-0.55/12 6} h5 {(Nd7) 0.01/1 3} 14. Bxc4 {
-0.24/12 6} Nd7 {0.01/1 5} 15. Bf4 {0.00/11 4} Bh6 {(Qf6) 0.01/1 5} 16. Bxh6 {
0.50/11 2} Rxh6 {(Nxe5) 0.01/1 3} 17. Bd3 {1.03/10 3} Nxe5 {0.01/1 2} 18. dxe5
{0.63/12 4} c5 {0.01/1 3} 19. Bxe4 {0.73/12 2} Bxe4 {(Qxd1) 0.01/1 3} 20. Nxc5
{0.73/12 1} Bc6 {(Qxd1) 0.01/1 3} 21. a3 {0.45/12 13} Qxh4 {0.01/1 1} 22. Qd6 {
0.99/9 1} Rc8 {(Bb5) 0.01/1 3} 23. axb4 {1.40/9 3} Bxg2 {(Qe7) 0.01/1 3} 24.
Qd7+ {-5.47/9 7} Kf8 {0.01/1 0} 25. Kxg2 {-7.45/10 2} Qh3+ {0.01/1 3} 0-1

[DomLeste]

Im going to test SP 32 bit version for CCRL as i only have Dual core duo not quad

Seems like its very strong

[krazyken]

Thinker seems to do well at very fast time controls. I have run 40 moves in 4 secs tournaments where Thinker is on par with Hiarcs 12.1 and Fruit 2.3.1. This TC is too fast for Shredder to handle though. More details and crosstable can be found here.

[Eraserheads]

Very impressive result! But for style-fanatics like me, the question remains: Has its style changed from 5.1e?

[Ovyron]

Very impressive result! But for style-fanatics like me, the question remains: Has its style changed from 5.1e?

Due to a bug in King Safety 5.2C was playing more boring chess than the 5.1 series but I can confirm that the attractiveness is back in 5.2D (I haven't checked 5.2E yet.)

[swami]

Would be interested in Thinker vs Hiarcs...

I wonder if there will be newer Shredder follow up to that FRC tourney?

[Marc MP]

1m+1s, D945 3.4 GHz
128 MB hash tables
single engines, ponder off
Gambit Suite 2008 by Jeroen Noomen
Fritz 11 GUI, Vista 32

Code: Select all

1   Thinker 5.2e-32 passive  +38/-29/=33 54.50%   54.5/100 --> +33 Elo
2   Shredder 11 SE(*)        +29/-38/=33 45.50%   45.5/100

An excellent result! Thinker had a great start with 7.0/9. I was using Wb2Uci, but did not notice any technical problems this time (except the missing eval+depth infos). Thinker doesn't ponder though, in Fritz with Wb2Uci. I know the old Wb. adapter but I noticed, Thinker starts to ponder but stops during ponder time. So I do not trust it, and preferred a ponder off match.

The result is a mix of the performance from the gambit perspective and from the opponent's perspective; each variation was played twice. Due to style elements, it is of course possible that Thinker scores extra good versus Shredder in these gambit type positions, and maybe Shredder would score better against Thinker in more calm openings. Although, Shredder has been described as an aggressive engine too, so I do not think that this makes a big difference.

Despite the very short time control, the conditions seemed ok to me: Shredder achieved typical depths of 10...12 plies during the middlegame.

*) The S11 SE version does not use tablebases nor Shredderbases in the search. CEGT has this version in the 40/4 list. 33 Elo higher we find engines like Rybka 1.1 and 1.2f 32-bit or Shredder 11 64-bit. They have it also in the 40/20 list.

[Event "Thinkertest (single)"]
[Site "COMPUTER-11"]
[Date "2008.08.09"]
[Round "32"]
[White "Shredder 11 SE"]
[Black "Thinker 5.2e-32 passive"]
[Result "0-1"]
[ECO "C27"]
[PlyCount "44"]
[TimeControl "60+1"]

1. e4 e5 2. Nc3 Nf6 3. Bc4 Nxe4 4. Qh5 Nd6 5. Bb3 Nc6 6. Nb5 g6 7. Qf3 f5 8.
Qd5 Qe7 9. Nxc7+ Kd8 10. Nxa8 b6 {Both last book move} 11. Nxb6 {2.29/11 7}
axb6 {0.01/1 1} 12. Ne2 {2.01/11 2} Bb7 {0.01/1 2} 13. Qf3 {2.05/12 4} Nd4 {
0.01/1 3} 14. Qh3 {2.16/12 6} f4 {(Nxe2) 0.01/1 3} 15. Nxd4 {2.19/10 1} exd4+ {
0.01/1 3} 16. Kf1 {-1.79/11 9} Bg7 {(Bh6) 0.01/1 3} 17. Qg4 {-1.51/11 3} Re8 {
(f3) 0.01/1 3} 18. Be6 {-4.33/11 12} Nf5 {0.01/1 3} 19. Qd1 {-4.33/11 2} Qxe6 {
0.01/1 3} 20. f3 {-4.43/11 2} d3 {(Ba6+) 0.01/1 3} 21. c3 {-9.48/10 4} Bf6 {
(Nh4) 0.01/1 3} 22. b3 {-9.86/11 7} Nd4 {0.01/1 3} 0-1

[Event "Thinkertest (single)"]
[Site "COMPUTER-11"]
[Date "2008.08.09"]
[Round "80"]
[White "Shredder 11 SE"]
[Black "Thinker 5.2e-32 passive"]
[Result "0-1"]
[ECO "D43"]
[PlyCount "50"]
[TimeControl "60+1"]

1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. Nc3 e6 5. Bg5 h6 6. Bh4 dxc4 7. e4 g5 8. Bg3 b5
9. Be2 Bb7 10. h4 g4 11. Ne5 {Both last book move} b4 {0.01/1 2} 12. Na4 {
-0.43/12 5} Nxe4 {0.01/1 3} 13. O-O {-0.55/12 6} h5 {(Nd7) 0.01/1 3} 14. Bxc4 {
-0.24/12 6} Nd7 {0.01/1 5} 15. Bf4 {0.00/11 4} Bh6 {(Qf6) 0.01/1 5} 16. Bxh6 {
0.50/11 2} Rxh6 {(Nxe5) 0.01/1 3} 17. Bd3 {1.03/10 3} Nxe5 {0.01/1 2} 18. dxe5
{0.63/12 4} c5 {0.01/1 3} 19. Bxe4 {0.73/12 2} Bxe4 {(Qxd1) 0.01/1 3} 20. Nxc5
{0.73/12 1} Bc6 {(Qxd1) 0.01/1 3} 21. a3 {0.45/12 13} Qxh4 {0.01/1 1} 22. Qd6 {
0.99/9 1} Rc8 {(Bb5) 0.01/1 3} 23. axb4 {1.40/9 3} Bxg2 {(Qe7) 0.01/1 3} 24.
Qd7+ {-5.47/9 7} Kf8 {0.01/1 0} 25. Kxg2 {-7.45/10 2} Qh3+ {0.01/1 3} 0-1

Hi Mike,

I tested Thinker Passive 5.2D (ponder off) at 4m+4s on a 2.4 Ghz pentium against Rybka 2.2n2 (both with 128 hash), and the result was (using Noomen's Gambit Suite): Rybka 2.2n2 64 - 36 Thinker Passive 5.2E.

[ArmyBridge]

"Although, Shredder has been described as an aggressive engine too, so I do not think that this makes a big difference. "
Hi Mike, I don´t think that S11 is agressive, S10 is agressive, I´ve play trough some games of S11 and for me, S11 is more positional than S10. Anyway it´s very interesting your results, I used to think that Shredder is very efective against agressive engines (especially vs Junior) so is amazing the Thinker result.
Regards