Author: Terry McCracken
Date: 21:55:45 07/27/01
Go up one level in this thread
On July 26, 2001 at 22:30:35, Robert Hyatt wrote: >On July 26, 2001 at 17:52:19, Jeroen van Dorp wrote: > >>It's about the time-trouble.... >> >>Forgive my ignorance, but with the same reasoning a program with the "ponder is >>always on-code" could decide to cut off calculations earlier because it thinks >>it has some calcluations done already -or will do - which in fact isn't the >>truth. >> >>In that case not *time trouble* but bigger chance of flawed analysis (because it >>allots itself too little time to think) would be the problem. >> >>IOW: a chess engine wouldn't use extra time with ponder=off because it also >>counts the non-existing pondering time during opponent's moves,so *no* (extra) >>chance of time trouble. >> >>But it *would* run into calculation problems because the time allocation for >>proper calculation is wrong. >> >>Where am I going wrong? >> >>J. > > >Nowhere. there are a hundred different ways ponder=off can confuse an engine >that wasn't thoroughly debugged in that mode. > >But it seems pointless to keep explaining this... I understand your reasoning Dr. Hyatt, and respect your opinion, you deserve it! A PHD and 30 years+ in computer programming! Just a question, why does Fritz GUI give a warning, "Do you want to run on only one cpu?" Not an exact quote. But I think Chessbase feels ponder=off on one cpu is the better option? Otherwise why is the question asked? (Default is Ponder=off.) I'm not saying that ponder=off is better or as good as ponder=on , but why does Chessbase give that impression? It appears they are suggesting ponder=off is better on one machine as it divides thinking/calculations, by 50% with ponder=on. Terry
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.