Author: Bob Durrett
Date: 09:42:39 01/14/04
Go up one level in this thread
On January 14, 2004 at 03:24:10, Jouni Uski wrote: >Many testers (specially in Germany) play their test games with x basic minutes + >y seconds increment. What is the benefit of this kind of hydrid level really? >I still prefer to play x moves in y minutes . Of course reason for increment is >to make time losses almost impossible, but this is wrong way I think. If engine >loses on time it's engines own fault and should punish for that end of >discussion! With x/n You can count tournament time by estimating around 70 moves >in each game. > >Alternatively You can of course play game in x minutes but still no increment >needed. > >Jouni If I were designing a chess-playing program, I would think long and hard about what KINDS of chess the program was to be used for. This is a matter of design being driven by the design objectives. Intended use sets the design objectives. A "general-purpose" chess-playing program would be intended for a wide range of types of chess. I think Fritz is like that. The decision to create a "general purpose" chess program implies compromises between competing intended usages. If a "black box," such as a chess-playing program, is to be tested to evaluate it's performance, then the performance must be evaluated in terms of how well it meets the intended purposes or uses. Typically, the intended purposes or uses are translated into more speicific design objectives prior to beginning the design process. Nowadays, both of the kinds of time limits are in common use, so a good chess-playing program should be able to play either kind well. A complete test of the chess-playing program would evaluate performance for each type of time limit. Incidentally, the impact of the hardware [and any external software used in real time] on performance must be taken into account. If the program is playing on a chess server, then the "hardware" includes the server as well, perhaps, as the internet. These are just generalities, but the "bottom line" is clear: Engines must be designed for a variety of time limits, and they should be tested for this variety. Bob D.
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.