Author: Christophe Theron
Date: 15:05:21 02/05/01
Go up one level in this thread
On February 05, 2001 at 15:40:21, Paul wrote:
>On February 05, 2001 at 12:39:23, Christophe Theron wrote:
>
>>On February 05, 2001 at 09:15:52, Paul wrote:
>>
>>>I've been looking through the archives a bit, and came across this mate that Dr.
>>>Dann posted a long time ago, it's not too easy, but not too difficult either for
>>>top programs (but mine has a hard time finding it :) ...
>>>
>>>2r3k1/5ppp/7r/Q7/3P1p2/1N3Pnq/PP3K1P/R5R1 b - -
>>
>>Well this time Tiger took 98s to find the mate in 10. :(
>>
>>Ne4+... (K6-2 450, 6Mb hash)
>>
>>
>> Christophe
>
>Well, don't be disappointed Christophe ... mine took 03:12 for a mate in 11 :)
>
>I've been changing so much in my search the last few days, sometimes it looks
>like it's drunk, haha! I just read that Genius 6.5 & CM7000 found a M10 in ~15
>seconds on a p2/400, and there also is a mate in 9 (proven by Chest).
>
>Ok, I have 2 questions for you, but if you don't want to answer them I
>understand completely, just can't resist. :)
>
>1. Is your QSearch relatively simple like mine, or do you extend/look at
>check(evasion)s etc in there like Ed seems to do?
My QSearch is much more complex than that.
>2. I'm going to convert my movegen to 16x12 any day now (like you have if I
>remember correctly). One thing I don't like is that some arrays get twice as
>big, like the history one and the ones for piece-hash-value-lookups. It's
>possible to prevent that by calculation, but do you have any recommendations?
You can have smaller arrays by first converting the 16x12 coordinate into a
0..63 coordinate. It just takes one array lookup, it not very expensive and very
cache-friendly.
>Wish we had palmtrees over here in Amsterdam :)
I understand! :)
Christophe
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.