Author: Dann Corbit
Date: 14:45:55 11/30/00
Go up one level in this thread
On November 30, 2000 at 17:33:17, Laurence Chen wrote: >On November 30, 2000 at 16:03:02, Garry Evans wrote: > >> >> Looks like already the k62-450 is hopelessly outdated, isn't it time for the >>ssdf to upgrade? I assume that most people have atleast a 700mhz computer these >>days. Or am I mistaken? >They would if they could afford, unless anyone can make hardware donation to the >SSDF. Actually, using the older hardware is much smarter than using new hardware. You would have to completely recalibrate all the programs, taking tens of thousands of hours to get an accurate figure. Besides, it is quite likely that the programs scale approximately linearly with hardware speed compared to each other, but that's another experiment. Consider: You have a magic opponent who has played literally millions of games. His ELO is known within 1/2 of a point. You have a steady stable of these opponents, all with the same rock-solid calibration. You have another list of opponents you can use. They might be able to beat Kasparov, but they might be 300 ELO below him. They are probably stronger than the first group, but you can't really even be sure of that. You really don't know all that well. Which pool of talent will you use to find out how strong a new engine is? If you are smart, or know anything about mathematics, you will use the first pool. For those who would like to use high speed hardware to calibrate chess programs -- have at it. There is nothing to stop you from doing so except, perhaps, the tens of thousands of hours of efforts between the start time and the time you have good, reliable numbers.
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.