Author: Enrique Irazoqui
Date: 04:08:19 11/12/00
Go up one level in this thread
On November 12, 2000 at 06:52:04, Uri Blass wrote: >I believe that some programs have clearly diminishing return in plies when other >programs do not suffer significantly from this problem. A quick and dirty way to comapre engines in this regard is checking in the SSDF list the increase in rating points when running on P90, P200MMX and AMD 450. Some programs have been tested with the 3 platforms, some others with 2. A case in point would be Hiarcs 7.32, relatively stronger on P200MMX than on the AMD 450. It is also known that H732 is comparatively better at blitz than at 40:2, for the same reason. >I suggest the following experiment to check it. > >1)Give chess programs 5 hours to calculate a fixed set of positions(I suggest >the position after 20 moves of both sides from the human-machine challange when >everyone can download the games at club kasparov). 3rr1k1/pp1nqp1p/2pb1np1/8/3N4/PP2P2P/1BQ1NPP1/R4RK1 b - - 0 22 That's an "average" middlegame position. You could do your experiment with it. Enrique >2)Write the depth after 5 seconds,after 5 minutes and after 5 hours > >3)Calculate the average depth after 5 seconds,the average depth after 5 minutes >and the average depth after 5 hours. > >I am interesting in comparing the results of different programs. >I am interesting to know results of programs that I do not have (Fritz6a or Deep >fritz,Nimzo8 and chessmaster8000,Zarkov,chess system,shredder4 and >chesssystemtal) in order to compare the results with programs that I >have(Gandalf4.32,Tiger13,Gambittiger,RebelCentury3,chessmaster6000,Hiarcs7.32,old >Fritz versions,Crafty,and Shredder5 that I am going to get in the future). > >I guess that I am going to try only part of the programs that I have so if >somebody is interested in checking them I have no problem with it but in this >case I ask him(her) to tell me because I do not want to do the same job again. > >Diminishing returns in plies does not prove problems at long time control >because the extensions and the evaluation may be more important but my guess is >that it suggest that there is a problem. > >I am going to use PIII450 for the experiment and people can use 3 seconds,3 >minutes and 3 hours if they believe their hardware is almost twice faster. > >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.