Author: Will Singleton
Date: 19:38:49 03/20/00
Go up one level in this thread
On March 20, 2000 at 22:02:35, Dann Corbit wrote: >On March 20, 2000 at 20:50:30, Will Singleton wrote: >[snip] >>I've been fooling with null-move a bit, disabling it when attacking around the >>king. Added lazy-eval, which sped things up a bit. Various bug fixes. What I >>really need to do is rewrite the whole thing, but no time for that. > >If you had all the time in the world to rewrite it, what changes would you make? If I had a few weeks to think about it, maybe I could give you a good answer. :) Offhand, I'd say that evaluation and search are two good areas for improvement. bsf, I see these new guys coming out with progs they started a couple of months ago, and getting 200-300 knps, and I guess it's due to the bitboard approach. So I'd try that. Then, reorganizing the eval into logical components to enable a better lazy eval, and to provide ways to further cut down on the time spent in the eval routine. Then I'd finally try to get the brainy bolt-on working, and get a real book (with learning), and egtb's, and do some cool endgame stuff, and then I'd port the whole thing to unix so I could do multi-processing, and then I want to write an analysis tool to look at my search trees and identify inconsistences, repeated stuff, errors, and unpromising lines. Then there's always the debugging, where you must spend days analyzing games, finding problems, runing test suites, etc. Lots to do Will
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.