Author: Bas Hamstra
Date: 06:25:54 01/13/03
Go up one level in this thread
On January 13, 2003 at 08:35:47, Sune Fischer wrote:
>On January 13, 2003 at 05:57:25, Bas Hamstra wrote:
>
>>On January 13, 2003 at 05:31:05, Uri Blass wrote:
>>
>>>On January 13, 2003 at 05:05:05, Bas Hamstra wrote:
>>>
>>>>>i hope you realize that there is a big contradiction between 'extensive pawn
>>>>>structure analysis', fast search, checks in qsearch, good move sorting and fast
>>>>>nps.
>>>>
>>>>Yes, probably. You start to wonder how Frans did it in Fritz5. But F5 proves
>>>>that is IS possible. At least checks/speed/sorting. Although sorting I'm not
>>>>sure of, with F5's extremely heavy dependency on hashsize.
>>>
>>>It is easy to show fast search all you need is printf("%d",node*10);
>>>
>>>I really do not care about nps when there are different definitions of nps( for
>>>example I do not count null move when Crafty does).
>>>I can have millions of nps if I count every legal move that I generate and not
>>>only every legal move that I make.
>>
>>Still it is easy to compare fairly. If you know Crafty's definition, change one
>>Nodes++ statement in Movei and you can compare with Crafty. In my case I know
>>how I count nodes (actually this reminds me of a statement of George W. Bush "I
>>know what I believe, and I believe that what I believe is right") so I can
>>compare with Tao (and Crafty because I count the same way).
>
>It's not that easy to compare because illegal moves made are counted too in
>Crafty AFAICT, they are not un-counted when they are un-made.
Ah I see, you mean in the qsearch. Yes, then Tao differs from Crafty since in
the qsearch I call GenOutOfCheck() (which does not guarantuee only legal moves
so actually I am "in between").
>The problem as I see it, is that this method just measures air, it's empty
>nodes.
>Suppose you want to see if the investment of doing InCheck is worth it, if you
>count illegal nodes all you will ever conclude is that it slows you down. On the
>other hand if you only count legal moves, then you might see a speed increase
>because you save a lot of doing and undoing moves that were illegal.
>
>Whether it is really worth it, depends on the speed of you make/umake and
>generate captures routines versus the InCheck and check evader routines.
>The proper way to find out is to count only legal moves, IMO.
Good point. On the other hand it might become clear that you search slightly
deeper when NOT doing the legality check in the qsearch. But in Tao differences
were very small, last time I tried.
Bas.
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.