Author: Ed Schröder
Date: 07:54:31 03/29/02
Go up one level in this thread
On March 29, 2002 at 09:18:30, Gian-Carlo Pascutto wrote: >On March 29, 2002 at 08:40:48, Ed Schröder wrote: > >>>I mean that gandalf is 5 times faster on A1200 when Fritz >>>may be only 3 times faster on A1200. >> >>Where does this myth come from? :) > >One very simple explanation is that all hardware isn't the same >for all programs. Right, but not in the 40% range Uri suggested. >For example, Fritz is probably a very compact program, designed >so most critical stuff fits directly into the cache of the >CPU that was most current when it was designed. > >For other programs, this may be very different. You posted >yourself about this and the relation to Rebels knowledge >parameter. It is possible that future hardware (with bigger >cache) allows Rebel to play with a bigger knowledge setting >at nearly no speed loss. Clearly, this would be an advantage >of Rebel to Fritz, as the latter would not use the extra cache >so much. While I agree with this (speaking for Rebel only now) it's hardly an issue for nowadays hardware. Rebel in the past clearly favored AMD (15-25%) vs Intel, nowadays the net gain is very little, a few percent based on a comparison between a P3-800 and an AMD-1400. In fact my next (future) PC will be probably Intel based again as I am fed up with the AMD baking ovens, but I am wandering off the subject. So I think the criticism is justified, but not in the high range Uri suggested, a 40% deviation which I consider a myth. Last but not least and similar to this discussion is the question: "Would the SSDF list have the same order when played on Intel processors?". The question is asking for troubles :) Ed >-- >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.