Author: Charles Worthington
Date: 10:25:51 02/23/03
Go up one level in this thread
On February 22, 2003 at 22:07:48, Jeremiah Penery wrote: >On February 22, 2003 at 20:30:53, Charles Worthington wrote: > >>On February 22, 2003 at 20:28:26, Charles Worthington wrote: >> >>>long. The notion that double speed equals 20 elo is totally wrong from my >>>observation. The highest dual is at 3111 and the highest single is lucky to be >>>at 2900...and even at 2900 he wont be there long if he plays all challengers. >> >> >>In fact if you place a dual 2400mp in a 100 game match with a single 2400xp with >>equal ratings to start with the dual will be a whole lot higher than 20 above >>the single when they are done. > >I think people who say speed doubling equals 50(?) ELO are talking about >controlled, long time control games under controlled settings. In blitz games >on the internet, where people use all kinds of hardware and software settings, >perhaps changing them every game, the rating pool isn't very stable. Blitz time >controls tend to exaggerate the differences in hardware speed, because of >diminishing returns of search depth with respect to time. I.e., it takes more >than twice as long to reach the next ply, so at some point the speed doubling >will no longer give you any extra usable search depth. This affect doesn't come >much into play in blitz time controls. Actually I never thought about it but it_would_be a fascinating experiment. Unfortunately it would be almost impossible to conduct the experiment scientifically without having both the single_and_dual here to test it. Also identical opening books, RAM, etc. That would be an expensive test. But I could run the test with my 3.06 xeons vs the 3.06 p4 but the workstation has a good deal more ram and a better chipset so it wouldnt be 100% scientific but close enough for all practical purposes i guess. I'll give it some though. Does anyone have a link where I can download the 2600.ctg to conduct the test with?
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.