Thinker 5.2B pas vs. Shredder 9 SE, 28-22

[Mike S.]

3m+1s, Pentium D945 3.4 GHz
256 MB hash, ponder off
S.Canbaz' Top-25 openings
(both engines without tbs. access)

Code: Select all

1   Thinker 5.2b-32 passive  +22/-16/=12 56.00%   28.0/50
2   Shredder 9 SE            +16/-22/=12 44.00%   22.0/50

This includes two games where Thinker lost because the engine made random moves, immediatly. In the previous match, there was one such game and two losses on time because the engine didn't start.

Previous result (from 5m+1s, rest identical conditions):

Code: Select all

1   Thinker 5.1e-32 passive  +21/-20/=9 51.00%   25.5/50 
2   Shredder 9 SE            +20/-21/=9 49.00%   24.5/50 

(The only strength difference between S9 and S9 SE is, that the SE version doesn't access tablebases during the search.)

Two short Thinker wins with black, where I had the impression that Shredder was outperformed tactically:

[Event "Thi52p-S9SE 3m+1s "]
[Site "COMPUTER-11"]
[Date "2008.07.21"]
[Round "6"]
[White "Shredder 9 SE"]
[Black "Thinker 5.2b-32 passive"]
[Result "0-1"]
[ECO "B33"]
[PlyCount "72"]

1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 e5 6. Ndb5 d6 7. Bg5 a6 8.
Na3 b5 9. Bxf6 gxf6 10. Nd5 Be6 11. Qh5 Rg8 12. Qxh7 Rg6 13. Qh4 Rh6 14. Qg3 f5
15. Qd3 fxe4 16. Qxe4 Rc8 17. c3 b4 18. Nb1 f5 19. Qd3 bxc3 20. bxc3 Qa5 21.
Ne3 f4 22. Nc2 Rb8 23. Qxa6 Qxa6 24. Bxa6 Bf5 25. Nb4 Nxb4 26. cxb4 d5 27. Be2
Bxb4+ 28. Nd2 d4 29. Bd1 e4 30. Bc2 Rc8 31. Kd1 Rxc2 32. Kxc2 e3+ 33. Kb3 Bxd2
34. Kc4 exf2 35. Kxd4 Be3+ 36. Kd5 Bd3 0-1

[Event "Thi52p-S9SE 3m+1s "]
[Site "COMPUTER-11"]
[Date "2008.07.21"]
[Round "32"]
[White "Shredder 9 SE"]
[Black "Thinker 5.2b-32 passive"]
[Result "0-1"]
[ECO "D43"]
[PlyCount "70"]

1. d4 d5 2. c4 c6 3. Nc3 Nf6 4. Nf3 e6 5. Bg5 h6 6. Bh4 dxc4 7. e4 g5 8. Bg3 b5
9. Qc2 Nh5 10. O-O-O Na6 11. Ne5 Nxg3 12. hxg3 Bg7 13. Kb1 Qb6 14. f4 c5 15. a4
Nb4 16. Qd2 O-O 17. dxc5 Qxc5 18. Ng4 Bb7 19. e5 Rfd8 20. Qxd8+ Rxd8 21. Rxd8+
Kh7 22. fxg5 h5 23. Be2 Kg6 24. Nf6 Bxf6 25. Bxh5+ Kf5 26. gxf6 Bxg2 27. Bg6+
Kg5 28. Bxf7 Bxh1 29. Bxe6 Qg1+ 30. Rd1 Qf2 31. Rxh1 Qc2+ 32. Ka1 Nd3 33. Rb1
b4 34. Ne4+ Kh6 35. Bd7 c3 0-1

[swami]

Thanks for the tests, Your test results gives me an impression that the latest Thinker might be about 30-50 elo stronger.

[Ovyron]

This includes two games where Thinker lost because the engine made random moves, immediatly.

Was this with an adapter?

[Mike S.]

Yes, it was with the Wb2Uci adapter, version from June 16th, 2005. But I am sure that it was not the adapter's fault: I have used it with several other WB. engines and none had that problem.

[Ovyron]

Yes, it was with the Wb2Uci adapter, version from June 16th, 2005. But I am sure that it was not the adapter's fault: I have used it with several other WB. engines and none had that problem.

But Thinker does not have any problems without the adapter, so it seems as if Thinker and the adapter hated each other.

Anyway, I'm not sure the loses due to using an adapter should be counted if they don't happen without it.

[Mike S.]

I'm not sure the loses due to using an adapter should be counted if they don't happen without it.

These results are not part of a rating list, but more like a "snap-shot" performance under certain conditions. Usually, I mention losses due to technical problems and losses on time separately but I prefer to keep them in the total result. It is debateable.

It were only 5 games from 100 games total, with such problems. If another engine would fail technically in 10 or more games from 50, I would cancel the test completely (at least the test with that configuration).

In addition to the usual configuration infos (time control, cpu...) I think I should mention the interface, and the adapter if it was in use, next time. Then, people can see better if the result may be representative for their own hard- and software, or not.

[CThinker]

Yes, it was with the Wb2Uci adapter, version from June 16th, 2005. But I am sure that it was not the adapter's fault: I have used it with several other WB. engines and none had that problem.

But Thinker does not have any problems without the adapter, so it seems as if Thinker and the adapter hated each other.

Anyway, I'm not sure the loses due to using an adapter should be counted if they don't happen without it.

There is definitely a mismatch between the wb2uci adapter and the Thinker engine. It looks like the two keep exchanging result info at the end of a game, and just before the next game starts, the adapter thinks that the engine has ended the new game. The adapter then decides on the moves instead of getting them from the engine.

I did a quick fix for this, by never again sending the result info, unless a new game has started.

This fix is now in the 4.2C version. This version has also fixed the bug in move validation of EP captures. As usual, you can download it from the ChessThinker page (www.geocities.com/thechessthinker).

Cheers...

[CThinker]

Yes, it was with the Wb2Uci adapter, version from June 16th, 2005. But I am sure that it was not the adapter's fault: I have used it with several other WB. engines and none had that problem.

But Thinker does not have any problems without the adapter, so it seems as if Thinker and the adapter hated each other.

Anyway, I'm not sure the loses due to using an adapter should be counted if they don't happen without it.

There is definitely a mismatch between the wb2uci adapter and the Thinker engine. It looks like the two keep exchanging result info at the end of a game, and just before the next game starts, the adapter thinks that the engine has ended the new game. The adapter then decides on the moves instead of getting them from the engine.

I did a quick fix for this, by never again sending the result info, unless a new game has started.

This fix is now in the 4.2C version. This version has also fixed the bug in move validation of EP captures. As usual, you can download it from the ChessThinker page (www.geocities.com/thechessthinker).

Cheers...

Slight correction...

It should be version 5.2C (and not 4.2C).

[Mike S.]

Thanks!

[Ovyron]

Thank you!