Author: Uri Blass
Date: 13:42:08 11/11/05
Go up one level in this thread
On November 11, 2005 at 16:32:10, Rajen Gupta wrote: >"Hi Uri: Some further explanations: still too less games / Junior 9 benefits >more than Fruit from ponder=on / pure accident. > >Regards >Kurt" > >Hi Kurt > >about 1-2 years ago when i had 2 identical computers next to each other at home, >i played a series of matches between df8, junior8 and shredder8 and chessmaster >9 personalities at various time controls with ponder off and on; the results >were vastly different with ponder off and on; fritz was strongest with ponder >off while shredder was weakest with ponder off the reverse weas true with ponder >on; cm 9k (strongest personality) though the weakest in both modes of testing >also lost more with ponder on; these reproducible but vastly different results >led me to believe that ponder on is the only way to judge engine strength as >that most closely mimics human play and optimally uses time. I see no reason to prefer one option. part of the programmers test mainly with ponder off and not with ponder on because they have not dual machine and testing on chess servers is not testing in controlled conditions because opponents change. I think that it may be interesting to see both ponder off rating and ponder on rating. Ignoring one option is against part of the programmers. I also think for the same reason that it is interesting to see both testing without books(CEGT) and testing with book(SSDF) claiming that it is better to ignore one option is again against part of the programmers. 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.