Author: Uri Blass
Date: 23:32:13 10/16/01
Go up one level in this thread
On October 17, 2001 at 01:09:35, Emil Vlasak wrote: >>1) Why does Fritz 6 improve more than Deep-Fritz for increasing frequencies? >>2) Are you using a linear extrapolation to calculate these values? >>3) It appears to be calibrated for Intel only. The fit should depend on the type >>of CPU. No? > >Im using a logarithmic extrapolation based only on SSDF list. >The source is available directly in MEC.HTM - you can >see it for example directly from IE. >DeepFritz was tested on two CPUs so it was calculated from two >values. Fritz is calculated from an awerage koeficient... > >The ELO ich cheaply depended on MHz and only slightly on CPU type, >hash, HDD... see http://web.quick.cz/EVCOMP/evcpu.htm#TestCPU > >Of course, it is changed with high speed TBird/P4, but this is not >calculated in my first version. Maybe there is a simple solution - >for new Athlons Palomino use the from AMD given pseudofrequency >rating. > >Emil All these calculations is based on wrong assumptions The speed of programs is not close to be proportional to the Mhz. I do not need to look at the second site to see it. It is known based on previous post of jouni aski that for Gandalf 1200Mhz of the ssdf is 5.3 times faster than the 450Mhz of the ssdf when for tiger 1200Mhz is only 2.67 times faster. I do not see a reason to believe that gandalf is going to earn more speed from mhz in the future. 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.