Author: Uri Blass
Date: 10:06:07 02/19/01
Go up one level in this thread
On February 19, 2001 at 12:27:35, Jeremiah Penery wrote:
>On February 19, 2001 at 04:20:51, Uri Blass wrote:
>
>>On February 18, 2001 at 15:37:30, Jeremiah Penery wrote:
>>
>>>On February 18, 2001 at 03:40:35, Tanya Deborah wrote:
>>>
>>>>On February 18, 2001 at 03:03:43, Jeremiah Penery wrote:
>>>>
>>>>>On February 18, 2001 at 01:22:38, Tanya Deborah wrote:
>>>>>>
>>>>>>Hi to all,
>>>>>>
>>>>>>This is a beautiful game by the Great Master Leonid Stein.
>>>>>>I am interested to know, How many programs can avoid the fatal mistake 21.Qxh1?
>>>>>>
>>>>>>Deep Fritz can avoid that!!!, Deep Fritz can see that Qxb3 is better than Qxh1.
>>>>>> Junior6 and Hiarcs7.32 and Fritz6 can´t avoid that...
>>>>>>
>>>>>>The position is very interesting, in this kind of position you can see some
>>>>>>computer weak points, (the machines sometimes have so much appetite, and can´t
>>>>>>see the great atack by White after move 21.
>>>>>>
>>>>>>Another question? Which program can find 12.Qb3! ???? (with a winning endgame)
>>>>>>- because after change Queens, White is much better. I think there are no
>>>>>>program that can find this move.
>>>
>>>>>>[Event "Moscu, 1966 -Spartakiada"]
>>>>>>[Date "1966.02.17"]
>>>>>>[Round "?"]
>>>>>>[White "Stein, Leonid"]
>>>>>>[Black "Birbrager Isaak"]
>>>>>>[Result "1-0"]
>>>>>>[ECO "B10"]
>>>>>>[PlyCount "43"]
>>>>>>[EventDate "2001.02.17"]
>>>>>>
>>>>>>1. e4 c6 2. d3 d5 3. Nd2 dxe4 4. dxe4 Nf6 5. Ngf3 Bg4 6. h3 Bh5 7. e5 Nd5 8. e6
>>>>>>f6 9. g4 Bg6 10. Nd4 Nc7 11. c3 Qd5 12. Qb3 12... Qxh1 $4 {A big mistake!,
>>>>>>Black should have change Queens. (Dxb3). Now with this Rook sacrifice, White
>>>>>>is winning! -(Stein.)}
>>>>>> 13. Qxb7 13... Kd8 {
>>>>>>A very dramatic and obligatory King move! - Black position is hopeless!} 14.
>>>>>>N2f3 Bd3 15. Bf4 $1 {A very fine move!} 15... Qxf1+ 16. Kd2 Qxf2+ 17. Kxd3 {
>>>>>>Black is lost
>>>>>
>>>>>The line of Crafty agreed with the analysis until here. I think white can only
>>>>>get a draw after Nba6 instead of Nxe6 in this position.
>>>>>[D]rn1k1b1r/pQn1p1pp/2p1Pp2/8/3N1BP1/2PK1N1P/PP3q2/R7 b - -
>>>
>>>>>What is the relevant line after 17. ...Nba6 for White to win?
>>>>
>>>>After Nba6 DEEP FRITZ see that White can win easy with :
>>>>
>>>>18. Bxc7+ Nxc7 19. Rd1 c5 20. Nb5 Nxe6 21. Qxa8+ Kd7 22. Kc4+ Nd4 23. Nbxd4 cxd4
>>>>24. Nxd4 e5 25. Qxc6+ Ke7 26. Qe6+ Kd8 27. Nc6+ Kc7 28. Rd7+ Kb6 29. Nd4+ Ad6
>>>>30. Qxd6 Ka5 31. Rxa7 MATE!! (DEEP FRITZ)
>>>
>>>Crafty did fail-low on Nba6 shortly after I posted the message - I just didn't
>>>let it search long enough before I posted. However, I didn't let it search long
>>>enough to resolve the fail-low, so I don't know what line it found. Thanks for
>>>the analysis. :)
>>
>>I suspect that the singular extensions that you use are counter productive.
>>
>>I think that you should compare the time that Crafty need to find the right
>>moves with singular extensions with the time that Crafty needs to find the right
>>move without them.
>
>I'm more interested in making Singular Extensions to work better in Crafty, but
>I've had no time to tune them at all. Hopefully I should be able to minimize
>the time loss but still keep most of the gains.
>
>>Crafty need less plies with singular extensions but I suspect that it needs more
>>time.
>>Crafty18.01 under chessbase needed only 48 seconds to see 0.00 evaluation for
>>Qxh1 on p800.
>
>How many NPS does it see on that machine? I usually get only about 100-150KN/s
>(with any version of Crafty). If you're getting, say, 500KN/s on that machine,
>then it's probably about 4x as fast as mine.
424 Kn/s in the case of the 48 seconds.
It is still faster than you singular extensions version based on your results
because your crafty needed more than 6 minutes when my 48 seconds are translated
to something like 3 minutes on your machine.
I used chessbase version but I guess that the chessbase version is not better
than the original one.
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.