Author: Dan Newman
Date: 23:45:37 05/03/00
Go up one level in this thread
On May 03, 2000 at 18:17:15, Daniel Clausen wrote: >Hi > >On May 03, 2000 at 10:20:14, Severi Salminen wrote: > >>Hi everybody! >> >>I would like to know how fast should my program reach (say) ply 8 from starting >>position, using AB, no hash tables, no extensions, no null move pruning. Just >>plain AB with ordering and very simple evaluation (material, mobility and >>developement). How many nodes/second should my program access? I'm asking this >>to see if my program has bad basic routines which I should improve before other >>things. I am using an old Celeron (which lacks some inner cache), 300Mhz. How >>fast are your programs? > >I don't know how much you did for your engine already, but in my engine I >couldn't care less about speed for the time being. I'd consider it a waste >of time to optimize eg the make-move part at the moment, because it will change >quite a lot in the future I think. Of course after a while these changes will >settle down a lil. This will be the time when I'll begin to optimize. >Just my opinion of course. > >Kind regards, > -sargon This is probably good advice, but I've never been able to follow it. I'm usually optimizing, tweeking, and so forth as I go along. I had about decided to give up this compulsive tweeking as useless when I suddenly came to the realization that it actually had some benefit--not in making the code faster--but in finding bugs. When you fiddle with code you have to look at it closely, more closely than you would if you were simply reading it. I recently discovered a long standing (and very well hidden) bug this way (which is what gave me this sudden insight :). -Dan.
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.