Author: Bernhard Bauer
Date: 04:31:26 12/06/99
Go up one level in this thread
On December 03, 1999 at 23:56:36, Dann Corbit wrote:
>On December 03, 1999 at 08:05:23, Bernhard Bauer wrote:
>
>>On December 03, 1999 at 01:27:03, Jouni Uski wrote:
>>
>>>As first program ever Nimzo got 30/30 in that test, when PC is AMD K6-2/450 and
>>>hash 40MB.
>>>Individual results in seconds (all positions kept at least 2 more plys):
>>>
>>> 1 2 3 4 5 6 7 8 9 10
>>>0 1 2 3 58 3 2 1 1 2 5
>>>1 1 1 573 1 4 180 1 1 3 2
>>>2 28 1 4 8 4 1 1 108 95 2
>>>
>>>Quite stunning!
>>>
>>>Jouni
>>>
>>>Test available at:
>>>http://www.geocities.com/CapeCanaveral/Launchpad/2640/pgn/tests/index.htm.
>>
>>Some years ago we had the results for this test (BT2630) of many programs.
>>That was interesting.
>>So I will give the results for Crafty too.
>>
>>Hardware: Intel 2x P3-450
>>Software: Crafty 17.3 24,8,8 Mb hash
>>
>> I 1 2 3 4 5 6 7 8 9 10
>>--I-----------------------------------------------------------
>>0 I 1 4 9 17 0 0 0 0 18 3
>>1 I 0 0 248 2 114 900 4 12 26 0
>>2 I 270 0 900 900 7 0 4 213 900 0
>>
>>Total time: 4552; Elo: 2478
>>
>>Remark:
>>Pos 15: e5 is found after 4.7 sec at ply 9, but fails low at ply 11 and
>> Crafty likes Rf6+ followed by e5 with a score of 2.0 at ply 14.
>> 1. Rf6+ Kg7 2. e5 is shown after 114 sec.
>>Pos 16: Crafty 17.2 finds 1. ... Nxg2 after 25:42 min.
>>
>>Does anybody disagree with 1.Rf6+ in pos 15?
>The C.A.P. record for this position (looks like a Rebel generated one) says:
>pv e5 dxe5 Rf6+ Kg7
>with a depth of 13 ply and a ce of 146.
>
>The alternate evaluation is:
>ce 173
>pv e5 dxe5 Rf6+ Kg7 Rd6 Qf7 Qxf7+ Kxf7 Rxh6 Bg4 Nf6 Nxc4 Nxh5 e4 Nf6
>and appears to be crafty generated.
>
>What does your crafty say if you let it run longer? Did you fix the sign bug on
>the character array? There is one other thing that needs attention in the 17.3
>sources also:
>This is clearly a bug. The macro Captured can never yield a negative number.
> Captured (move) != -king &&
>valid.c(206) : Warning 650: Constant out of range for operator '!='
>
>Here is the actual definition:
>CHESS.H ( 1058): #define Captured(a) (((a)>>15)&7)
>
>If you have fixed both problems, what answer do you get with your crafty when
>allowed to run a long time on the position?
Crafty says:
14 12:53 2.05 1. Rf6+ Kg7 2. e5 dxe5 3. Rd6 Qf7 4.
Qxf7+ Kxf7 5. Rxh6 Nxc4 6. Rxh5 e4
7. Rh7+ Ke6 8. Nf4+ Ke5 9. Rxa7 Bg4
10. Rc7
No, I havn't changed anything.
What should I change exactly?
Kind regards
Bernhard
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.