Author: Uri Blass
Date: 16:13:28 02/14/02
Go up one level in this thread
On February 14, 2002 at 18:23:56, Gian-Carlo Pascutto wrote: >On February 14, 2002 at 18:10:53, Roy Eassa wrote: > >>A Go program that can run on an original IBM PC at 4.77 MHz would necessarily be >>at least a little stronger if it ran on an Athlon 2000+, no? The only way I can >>see that not being the case is if there is no time limit per move (or per game >>or per set of moves) that is of any constraint to any decent algorithm. [Even >>doing only a static position evaluation should take some non-trivial number of >>CPU cycles to achieve a given level of strength.] > >For your example, I don't see why it should. If it can play on the XT >in the required contraints, the _same program_ likely will gain very little >or no strength moving to an Athlon. > >You could argue the reverse and say that going from the Athlon to the XT >will lower the strength because it would exceed time and lose all of its >matches. > >With similar reasoning you can probably argue that chess is solvable >in O(1) time, that superlinear parallel speedups are possible, that >chess is not a finite game and quite a bit of things about Deep Blue too, >even. > >But I'm not going to reply. > >-- >GCP In every finite game when the rules are clear better hardware or more time can help If I have a program A that almost does not use time then it is easy to develop a program B that is going to win against A. B will try every legal move and continue by playing A against itself. B is going to choose the move that leaded to the best result against A. It is clear that B is going to beat A in a match and I cannot believe that B is not going to be at least slightly better than A against other programs. I can believe that software is very important in go and I can believe that the best program can beat the second best program even when the hardware of the second best program is 100 times faster but I cannot believe that better hardware cannot help without a design of the programmer not to use time intentionally. 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.