Author: José de Jesús García Ruvalcaba
Date: 08:39:53 06/26/00
Go up one level in this thread
On June 24, 2000 at 15:21:44, blass uri wrote: >On June 23, 2000 at 16:40:10, José de Jesús García Ruvalcaba wrote: > >>On June 23, 2000 at 14:05:59, Osorio Meirelles wrote: >> >>> >>>Any good estimates ? I wonder how much better the chess programs are really >>>improoving, if we put all of them at the same speed, so we can see the actual >>>improovement due to better algorithms . >> >> The improvement is not only due to better algorithms, but also due to >>algorithms adapted to faster hardware. >> If you run the new programs versus very old on very old hardware, I think the >>old programs have a chance to win (it is posible that the new ones lose all >>their games on time). > >They can lose on time only in blitz games but people are interested in 40/120 >games(the ssdf list is only about these games). > >If the hardware is 40 times faster than 2 hours/40 moves is the same as 3 >minutes/40 moves with the fast hardware and programs usually do not lose games >on time when I tell them 3 minutes/40 moves. > >Some program like Hiarcs7.32 or Crafty will never lose on time even at 1 >minute/game. > >Uri I was not thinking about forty times slower, but of the time when a full-width 3-ply alpha-beta search took several hours. But even at 40 times slower, some current programs which assume powerful hardware (crafty comes to mind) would have a hard time against programs tuned for that hardware. José.
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.