Author: Uri Blass
Date: 01:32:04 03/24/03
Go up one level in this thread
On March 24, 2003 at 04:15:45, Uri Blass wrote:
>On March 24, 2003 at 03:19:41, Will Singleton wrote:
>
>>On March 24, 2003 at 02:52:40, Will Singleton wrote:
>>
>>>On March 24, 2003 at 02:25:33, Bernhard Bauer wrote:
>>>
>>>>On March 24, 2003 at 02:07:31, Will Singleton wrote:
>>>>
>>>>>Position from Baron – Amateur bm Qxc2
>>>>>[D]2rr3k/pb3ppp/4p3/4P3/1n1B3R/1P4P1/P1q4P/1Q1RKB2 w - - 0 30
>>>>>
>>>>>For some reason, Ruffian and Crafty have a very tough time with this one. I
>>>>>wonder whether this could be due to some similarity between them.
>
>By this logic public movei that probably need more time on your machine is also
>similiar to Crafty.
>
>
>I think that a lot of engines are slower than Ruffian(probably most of the free
>engines).
>
> A problem in
>>>>>repetition detection, perhaps? On my machine, amd 1.6ghz:
>>>>>
>>>>> Fail-low solution
>>>>>Yace 0:15 0:20
>>>>>CM 0:15 0:30
>>>>>Shredder 0:35 0:45
>>>>>Amateur 0:20 0:50
>>>>>Ruffian 2:34 3:54
>>>>>Crafty 5:42 > 9 minutes
>>>>>
>>>>>---------------------------------
>>>>>[Event "Computer chess game"]
>>>>>[Site "MICRON"]
>>>>>[Date "2003.03.23"]
>>>>>[Round "1"]
>>>>>[White "The Baron 0.99.5a"]
>>>>>[Black "Amateur 2.44"]
>>>>>[Result "0-1"]
>>>>>[TimeControl "40/600"]
>>>>>
>>>>>1. d4 Nf6 2. c4 e6 3. Nf3 d5 4. g3 Be7 5. Bg2 O-O 6. Qc2 c6 7. O-O Nbd7 8.
>>>>>Nbd2 b6 9. e4 Bb7 10. e5 Ne8 11. b3 Rc8 12. Bb2 c5 13. Rac1 cxd4 14. Nxd4
>>>>>Kh8 15. Rfe1 Nc7 16. cxd5 Nxd5 17. Qb1 Nc5 18. Red1 Qd7 19. Ne4 Rfd8 20.
>>>>>Ne2 Qb5 21. Nd4 Qa6 22. Nxc5 Bxc5 23. Bf1 Qa5 24. Rc4 b5 25. Nxb5 Qxb5 26.
>>>>>Rh4 Bxf2+ 27. Kxf2 Qc5+ 28. Bd4 Qc2+ 29. Ke1 Nb4 30. Bxa7 Qc3+ 31. Kf2 Qf3+
>>>>>32. Kg1 Qh1+ 33. Kf2 Qf3+ 34. Kg1 h5 35. Rd6 g5 36. Rh3 Rxd6 37. exd6 Rc2
>>>>>38. Qxc2 Nxc2 39. d7 Qd5 40. Bb6 Qxd7 41. Rxh5+ Kg7 42. Rxg5+ Kf6 43. Rb5
>>>>>Qd1 44. a3 Bc6 45. Rb4 Nxb4 46. g4 Nd3 47. Bd4+ Ke7 48. Bc5+ Nxc5 49. h4
>>>>>{Baron resigns} 0-1
>>>>
>>>>Crafty 19.3 needs 28 sec on an AMD 1,8 GHz. So I can't see a very tough time.
>>>>Regards
>>>>Bernhard
>>>
>>>Very interesting. I was using Crafty 19.1. I just ran the position on 19.2 and
>>>19.3, confirming your results with 19.3. 19.2, however, behaves likes 19.1,
>>>taking several minutes. I'm off to check the source changes.
>>>
>>
>>More interesting. 19.3 has rewritten the repetition code. As I pointed out in
>>the original post, a problem with repetition detection might cause difficulty
>>with this position. What do you know.
>>
>>Will
>
>I do not understand what it has to do with repetition
>
>Movei0.0799 has problems in this position when Movei0.08031 can find it clearly
>faster(slightly more than 2 minutes on p850 for 0.08031 and more than 20 minutes
>for 0.0799 but I did not save exact solution times and it is not important
>because I did other things during the analysis but not things that made the
>program significantly slower based on looking at the nps).
>
>I did not change repetition detection code but I added evaluation
>of king safety and did a lot of other changes in the search.
>
>I do not see what repetition has to do with it because the evaluation is
>negative before the fail low and after the fail low(with king safety evaluation
>it is easier to find Qxc2 because with king safety evaluation it is clear that
>white need to trade queeens to prevent problems with king safety).
>
>Can you give the repetition line that give problems to programs to see the fail
>low?
>
>Uri
I can add that the main problem is null move pruning.
It is not easy to see the threat
after a line like Bxa7 Qc3+ Kf2 Qf3+ Kg1 h6
King safety evaluation may help not to prune the wrong lines and I guess that
minimal king safety evaluation is the reason that latest movei can find it
relatively faster(not as fast as amateur but at least significantly faster than
ruffian).
Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.