Author: Uri Blass
Date: 05:33:49 05/30/04
Go up one level in this thread
On May 30, 2004 at 08:13:36, Sune Fischer wrote: >On May 30, 2004 at 06:35:11, Uri Blass wrote: > >>On May 30, 2004 at 06:05:08, Sune Fischer wrote: >> >>>On May 30, 2004 at 05:35:31, Uri Blass wrote: >>> >>>>>I am doing mobility, but what has that got to do with move generation? >>>>> >>>>>-S. >>>> >>>>I do mobility by counting number of legal moves. >>> >>>My mobility is not so simple, I do different things for the different pieces. >>> >>>>move_generation is clearly relevant. >>> >>>I don't see any reason to generate moves, I think that's not what you need in an >>>eval. >> >>You do not need to remember the move but giving score for different moves can be >>relevant. >> >> >>> >>>>I will probably replace it in the future by better mobility but I plan to >>>>continue using similiar functions. >>> >>>This could be slowing you down quite a bit, should be worth experimenting to see >>>if you can find something cheaper which is just as good or perhaps better. IMHO. >> >>cheaper is not so simple because it also means rewriting my qsearch(today movei >>look for captures in the list of the legal moves so only changing the evaluation >>will not to things faster). >> >>It needs the number of legal moves for evaluation so only changing the qsearch >>will also not make it faster. >> >>I think to concentrate first on better and only later to think about cheaper. > >I don't think you can seperate it like that, a lot of things would obviously be >better from an "every node must be evaluated as accurately as possible" >viewpoint, e.g. probing TBs at the leaf. Of course but here I cannot do it cheaper without changing my qsearch code and I believe that it is possible to do it better when I do not need to change my qsearch code by evaluating every piece differently(something that I do not do today). Uri > >The real question is always if it is worth the cost. :) > >-S.
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.