Author: ed
Date: 06:02:26 10/21/98
Go up one level in this thread
On October 20, 1998 at 23:59:56, Bruce Moreland wrote: > >On October 20, 1998 at 21:48:04, ed wrote: > >>I dunno exactly how fast is dual 266, but according to bob, it should be .7 x >>CPU speed, so that is .7 x (266 x 2) so, that is about 373 mz. Even at same mz, >>PII is faster than a Cyrix for 32 bit instructions (~ 10% I guesstimate), so I >>would say that epita is about 20% faster than nudode. >> >>8-12 is the final score in favor of epita, but if u take into acct, of speed >>differential, the score should be close to even... Should have read "closer" to even. Btw, a 60% score is only worth ~ 60 points. Need 66% score for 100 points. But in any case, i agree with what bob said, a small delta in CPU speed between comp-comp games can mean a *huge* diff in final result, if not huge, certainly much more so than what one would expect. 60 elo+ performance points for higher rated comp2 in a comp1-comp2 game might translate to only 30+ elo for comp2 in a comp2-human vs comp1-human. Any takers? :) BTW, a small clarification, CometA96 seems to play better in middle game than crafty unless it falls for some tactical shots. It more likely than not almost always gets out of the opening with a + score. (dunno, if this is due to comptempt factor). >A 20% speed increase is worth a hundred rating points? > >bruce
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.