Author: Robert Hyatt
Date: 21:02:21 01/17/01
Go up one level in this thread
On January 17, 2001 at 11:10:13, Kim Roper Jensen wrote: >On January 15, 2001 at 21:00:42, Robert Hyatt wrote: > >>On January 15, 2001 at 18:13:00, Bas Hamstra wrote: >> >>>On January 15, 2001 at 11:25:53, Robert Hyatt wrote: >>> >>>>On January 15, 2001 at 10:45:21, Kim Roper Jensen wrote: >>>> >>>>>Hi >>>>> >>>>>I just wondered, If u have a slow processor how would u program it to play chess >>>>>?? >>>>> >>>>>I mean will u try to create a big eval and hope it selects "good & natural" >>>>>moves and have small depths, or do u try make a small eval and hope u can >>>>>calculate to a reasonable depth ???? >>>>> >>>>>Or is there no difference in program design, when u program a fast or slow >>>>>processor ?? >>>>> >>>>>With regards and thamks in advance >>>>>Kim >>>> >>>> >>>>There is definitely a "balance" point between evaluation and search. If you >>>>slow the machine down enough so that the big eval causes the search to be >>>>unable to reach a reasonable depth, then you have a positionally smart but >>>>tactically stupid program. The inverse can happen as easily. Striking a >>>>good balance between smarts and tactics is a big issue... >>> >>>Question is: is that balance different for slow cpu's? And question 2 is: are >>>you going to play fast cpu's? That extremely hard to compete anyway. For a game >>>between 2 very slow processors I would think the balance would more go to less >>>smart, more tactical, faster, than for fast processors. Stuff it with extensions >>>and make a Genius. For fast processors that doesn't work. >>> >>>Bas. >> >> >>I believe that the cpu speed makes a difference. I believe that it is >>possible (and probable) that if you play program A vs B, where A is designed >>and tested to run on a 200mhz processor, and B is designed on a 1ghz processor, >>that if both use 200 mhz processors, A will win, and if both use 1ghz >>processors, B will win. > >This is excaxtly what i mean, how will the implementation/design of the engine >differ from each other ???? > >> >>I had a similar problem in testing/developing Cray Blitz on a vax, but running >>it on a Cray. It was a real problem. I would characterize it as a set of trade-offs. IE I tried singular extensions many years ago on very slow hardware. It did poorly. I finally tried it again in the early 90's and it did fine on the Cray. But it did poorly on the VAX type machines. There are other examples of working good on fast hardware but poorly on slow hardware. And I can give one for the inverse if you want...
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.