Author: Russell Reagan
Date: 17:35:05 07/01/04
Go up one level in this thread
On July 01, 2004 at 11:07:29, Dan Honeycutt wrote: >On July 01, 2004 at 05:14:13, Fabien Letouzey wrote: > >>On June 30, 2004 at 23:22:12, Dan Honeycutt wrote: >> >>>Counterpoint: >> >>>http://f11.parsimony.net/forum16635/messages/67157.htm >> >>>Dan H. >> >>Hi Dan, >> >>I was not talking about computing them incrementally. >> >>Fabien. > >I took your comment to mean they are worth having even if they slow you down >when you first put them in. As your program progresses and you make better use >of them you regain the lost speed plus dividends. > >Best Regards >Dan H. That may have been what he meant, but that doesn't mean it always works that way. I don't think it is a clear cut situation. For one engine they might help as you described. For another they wouldn't help at all. Crafty, for instance, spends 1% of its time in AttacksTo(), and 4% of its time in Attacked(). The "attacks from" are in macro form, so they do not show up on a profile, but generating moves totals 8% so generating those attacks is surely less than 4% (probably less). Even if attack tables were completely free, you could not even get a 10% speedup (and attack tables are certainly not free). Attack tables will surely slow you down at least 10%, so there is really no point in trying them in Crafty. Even if I could get a 10% speedup, I don't know that the added complexity is worth it.
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.