Author: Robert Hyatt
Date: 18:24:34 07/30/01
Go up one level in this thread
On July 30, 2001 at 01:57:43, Miguel A. Ballicora wrote: >On July 29, 2001 at 09:50:23, Robert Hyatt wrote: > >>On July 29, 2001 at 06:51:54, Gian-Carlo Pascutto wrote: >> >>>On July 28, 2001 at 23:42:00, Robert Hyatt wrote: >>> >>> >>>>My conclusion... ponder=off was worse for crafty as its time management was >>>>not well designed nor tested for this mode. >>> >>>A simple (about 2 lines) fix would be to reduce 'usage' a >>>bit if pondering is off. >> >>It already does this. But the problem is that it is not tested much at >>all. That was the point of my original discussion about ponder=off matches >>using Crafty. One of the things that makes this program so dangerous at present >>(against humans) is the number of games it plays. It has played enough that I >>don't worry (as I did with Cray Blitz and predecessors) that it might well crash >>or misbehave in the infrequent games we played. >> >>The more a program is tested, the more reliable it will be. But for ponder=off, >>it is simply not tested... >> >> >> >> >>> >>>Easy to implement, fixes the most obvious problems, and it's >>>logical to have this. >>> >>>On Linux you can run with two engines with pondering on, but >>>I don't trust Windows much for that. Even more because some >>>engines (like Gerbil) lower their priorities. Running on of >>>them in this kind of match would be very unfair. >> >>Yes it would. That is the reason for my _second_ observation. one-machine >>matches are useless. There are simply too many variables in how the programs >>behave (or mis-behave). > >They are not useless, it all depends what your goals are. > >Regards, >Miguel No... They burn cpu cycles. Make sure your machine is working normally. Ditto for the disk drive, etc. But as far as testing which program is better in "normal mode" single-cpu matches are pointless. About as pointless as a single-bicycle bicycle race... > >> >> >> >> >>> >>>-- >>>GCP
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.