Author: Robert Hyatt
Date: 07:14:35 07/01/01
Go up one level in this thread
On July 01, 2001 at 02:11:32, Kurt Utzinger wrote: >On June 30, 2001 at 23:02:52, Robert Hyatt wrote: > >>On June 30, 2001 at 17:17:41, Fernando Villegas wrote: >> >>>Hi: >>>If you look at the ganme Adams won to Pocket Fritz, you will concur that we are >>>facing a program that, at least in this game, left the human side to kill him as >>>a child. The most elemental king attack I have seen in years and nevertheles >>>went OK. Blacks messing around in the queen side with no idea what to do as >>>amads prepared the knife. Not even elemental defensives moves to prevent what >>>was clearly coming with the obvious machine bishop-queen in the diagonal. >>>Sorry, but I have a Fidelity old piece of plastic that plays better than that... >>>Fernando >> >> >>You sound surprised. If you take a program that plays well at 1000K nodes >>per second, and run it on hardware where it can only search 1K nodes per >>second, what would you logically expect to happen? It will play with 1/1000th >>the skill (or even worse). > >In my opinion the whole Pocket Fritz concept is wrong. Why not use a more pure >tactical program instead of SMK's Shredder that has too much knowledge to run on >such a low hardware. >Kurt The real problem is hardware. A program is designed around a particular speed. If it goes slower than expected, it will probably play significantly worse as the program was designed so that the search would handle some tactics, and the evaluation handles what is left. If the hardware is too slow, the search doesn't hold up its end of things and the program is much worse than expected. If the hardware is much faster than expected, the search can duplicate part of what the evaluation does, also leading to problems. (this happened to me in the case of Cray Blitz in 1985/1986).
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.