Author: Thorsten Czub
Date: 17:22:37 01/20/04
Go up one level in this thread
On January 20, 2004 at 19:20:08, Paul Doire wrote: >Yes I have old hardware but the results are relative, of course they are reative. but relative to the hardware i do use 5 minutes on a 400 mhz machine is arround 1'40" blitz on my slow 1200 machines. >they all ran on it, >the same hardware. this is right, but it is not a point . let's give 5 cars the same engines, but different gearboxes. although all 5 cars have the same PS and the same engines, they will not speed up in the same way due to different gearboxes. When you now fotograph the ranking after 140 meters, there will be a winner, but another car will be the winner after 500 meters. chess programs behave like different gearboxes. some chess programs are starting very fast. others increase their strength after a while. so some programs have their peak in fast games, others in slow games. if you use a time control that is so slow that it is not measuring the strength but the velocity of the search algorithm, you do not measure the chess programs, but the search of the programs. IMO 1'40" blitz games on a 1200 mhz machine is TOO fast, especially if you do NOT play with an increment so that the computers have enough time. your games will lose sense in the moment the programs have only a few seconds left. your games will change into random or the results of the games will often change when time control is near. there is IMO not much time to play if you have only 1'40" on 1200 (or 5' on a 400 mhz). >It is possible that some engines are better optimized to take >advantage of fast hardware, this is true. your results could be completely different when using 20 minutes games, or even more different when using 40/40 or 40/120. >But how much would that have helped Hiarcs relative to other >modern engines also? I just don't know. I stand by my numbers for that answer. you do believe that the strength of chess programs increases linear. so you believe that hardware has a relative effect. but your assumption that the strength of chess programs increases linear is not true. the strength of a chess program increases linear in a range that makes sense. out of this range, the strength of different chess programs behaves completely different. if you drive with a car into a curve, the forces that lean the car will increase within a range, but suddenly, the forces make the car fall out of the track and you have an accident. that is because within a range everything is ok, but out of this range the behaviour is different. IMO 1'40" is out of range.
This page took 0.01 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.