Please read again. We were discussing this as a "general optimization". Which would suggest that it is a good idea for _any_ program. I can give specific quotes if needed. And in general, doing things "up front" is a bad idea in the context of alpha/beta, where delaying something for a bit may well turn in to delaying it forever.mcostalba wrote:In our case "Many" does not refer to any engine we are talking about in this threadbob wrote: Not necessarily. Many generate _all_ moves and then first pick out just the captures. You do not need to score non-captures yet as quite often a capture is enough to cause a quick cutoff.
BTW, I have to admit you are very good at generating "escape" moves
What makes Rybka so strong?
Moderators: hgm, Rebel, chrisw
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: What makes Rybka so strong?
-
- Posts: 12541
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: What makes Rybka so strong?
I think that material imbalance in the evaluation (while not a revolution) was a uniquely Rybka idea.