J. Varsoke a écrit :
> My primary use of Scid is to analyse my opening play. I have a very
> large database (3mil?) based on a converted ChessBase Mega + TWIC. I'm
> sure a lot of you have a similar setup.
>
> I usually open Mega, then open the Tree view. Hit the lock toggle.
> Then I go over to my game.
>
> I look at the win/loss and frequence of play for different moves while I
> move through the opening.
>
> What I'd like to see for Scid (and perhaps this is already partially a
> feature) is to be able to save the cache from the entire Big-Base as a
> file that I can load next time w/o having to generate the tree again.
>
> Instead of the tree being calculated to a certain number of moves per
> line -- it would be nice if a threshold of number of games played in
> that position; postitions that have only been played 10 times don't have
> much statistical merit.
>
> And really, I don't mind if it takes a week to generate this opening
> cache -- but I'd like to do it just once and load it after that.
>
> I know this is approximately what we've been talking about for the last
> week. I just thought I'd throw in my two-cents, and hopefully give
> Pascal some ideas.
>
> -j
>
Basically what you ask for is to cache most played moves, which
corresponds to main opening lines. This is why I tried 2000 positions
from ECO file, and the time to fill the cache is evaluated to several
weeks, and will still be useless because cache has a limited size
(currently 250 positions, next will be 1000). So the only solution is to
use a real book generated with polyglot (this is what I did with my own
3 M games base) along with a Tree window. I fill the cache with my own
bases containing my (small) opening repertoire.
For those willing to try the parameter to change is
const uint SCID_TreeCacheSize = 250;
in file tkscid.h
You can enter 20000 for example, I see no drawback in this except that
you'll end with a .stc file of about 400 MB (slightly bigger than the
base files). Note that if this parameter is changed, the cache file must
be completely regenerated.
But the time taken to fill such a big cache is deterrent for me.
Pascal
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Scid-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scid-users