Author: Uri Blass
Date: 09:07:55 03/23/02
Go up one level in this thread
On March 23, 2002 at 10:45:29, James T. Walker wrote: >On March 23, 2002 at 07:52:18, Gian-Carlo Pascutto wrote: > >>On March 22, 2002 at 18:32:56, James T. Walker wrote: >> >>>CSTal II hits about 30-35K/nps on my AMD 1.4G machine. On a K6-3-450 it usually >>>hit around 10K. Who knows when the hardware gets this thing up to about 1M/nps >>>it may be the strongest thing around. :=) Don't throw it away yet. >> >>I see absolutely no reason to assume it will play better compared to >>other programs when the hardware gets faster. >> >>In fact, people have said the exact same thing as you 2-3 years ago, >>and I only see CSTal getting worse compared to new programs, not better. >> >>-- >>GCP > >Hey GCP, lighten up! Can't you see the :-) ?? CSTal-2(.03) has some serious >problems which could have been fixed by the Genius if he was as smart as he >thought he was. It didn't happen. He gave up. I guess it was more than he >could handle. I know that chris wasted time on programming other games and I tend to believe that he could improve chess system tal if that was his target. I remember that chess system tal has some bugs and it is known that sometimes only removing bugs is responsible for a signigficant improvement. Sometimes the search baloons and CSTal can't handle the tactics. >It needed a complete re-write to speed up it's search. It is possible that a complete re-write can help but I believe that chess system tal waste most of it's time on the evaluation so in order to speed the search without changing the program chris needs to calculate the same evaluation faster. I tend to believe that it is possible to get a significant speed improvement but not easy. My guess is that if the quality of the programmers is the same then usually the more complicated program has more potential for speed improvement. Uri
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.