Author: Robert Hyatt
Date: 10:12:38 11/27/02
Go up one level in this thread
On November 27, 2002 at 13:02:19, Bob Durrett wrote: >On November 27, 2002 at 12:01:24, Robert Hyatt wrote: > >>On November 27, 2002 at 09:49:55, Bob Durrett wrote: >> >>> >>>Disclaimer: I do not know the specifics of how position evaluation is >>>accomplished in the top chess engines. >>> >>> >>>From what I have read here, position evaluation is typically accomplished with a >>>"function." This function appears to be fairly simple. Otherwise it would be >>>called a position evaluation "subprogram." Since a large number of positions >>>are evaluated, the evaluation must be very quick and hence simple. >>> >>>Less processor time is available for searching if the amount of processor time >>>used for position evaluation is increased. In an extreme example, one might >>>devote 90% of the processor time to position evaluation. I do not know what >>>this percentage is in practice. If a huge number of positions are evaluated, >>>then the total amount of processor time devoted to evaluation might be large. >>> >>>It seems that there is a tradeoff between complexity of position evaluation and >>>the number of positions evaluated. My impression is that current engines devote >>>only a very small percentage of the processor time to evaluation of any given >>>position. >>> >>>Devoting a large amount of processor time to position evaluation of each >>>individual position would make sense, or so it seems, only if the evaluation >>>were very very good. Essentially, this would mean evaluating only a relatively >>>few positions but evaluating them extremely well, using a complex position >>>evaluation subprogram. >>> >>>In theory, a perfect evaluation subprogram would have to evaluate only one >>>position for each move made. Maybe someone would say that chess engines do just >>>that. They "evaluate" the position after the move was made. They just do it >>>using search algorithms. >>> >>> >>>So, this leads to the following question: >>> >>>Has anybody yet explored the option of using complex evaluation subprograms >>>[which do not rely primarily on the use of search algorithms] for position >>>evaluation? >>> >>>Bob D. >> >> >>You definition of "simple" is very vague. For example, is Crafty "simple" or >>"complex"?? > >I was definitely NOT saying Crafty is simple!!!! I was referring to the >position evaluation code only. Also, I meant simple in a relative sense, when >compared to more complex evaluation, or to the overall chess engine program. I understood you to mean that. Since my evaluation takes over 50% of the total time that is used, it is far from "simple"... I'd call one "simple" it it is in the 10-15-20% range at worst... and even that would not be just a couple of hundred lines... > >>Whichever you choose, Crafty spends over 50% of the total time used for >>selecting a move in executing the evaluation code. > >********************************************** > >That is very interesting information! It would be nice to have similar >information for other chess engines. [Hint.] > >********************************************** > >To me, the gyrations a chess engine has to go through [alpha/beta and such], >other than execution of the evaluation "function," seem more like overhead. Of >course, this is merely my perception. Hard-core alpha/beta devotees would >probably take exception to this characterization. : ) > >>I have a hard time calling 50% of the total >>time "simple" myself, > >Well, there are two things being discussed here: (a) percentage of total >processor time used to perform position evaluation, and (b) complexity of the >evaluation versus the number of positions evaluated per unit of time. > >>since this is about 4,000 lines of code, roughly... > >Are all 4000 lines of this code within a single compact "function"? [Or is the >code somehow spread out here and there, willy nilly, throughout the Crafty >code?] It is almost all in evaluate.c... there are a few bits and pieces stuck over in utility.c as those are functions that can be used for evaluation _and_ other things... > >It would be easy to feel that a 4000 line evaluation "function" might not be >"simple," but when taken in the context that you have 51,000 lines of code in >Crafty, 4000 doesn't sound like much at all!!!!!!!! > >Anyway, my questions are driven by my perception that the evaluation function >size/complexity would be a critical parameter in chess engine design, at least >for the types of chess engines that do a lot of searching. > >I really would like to know about the tradeoffs that determine: >(a) the number of lines of position evaluation code as a percentage of the total >number of lines of code of the chess engine, and >(b) the percentage of total processor time to devote to position evaluation. > >How does the chess engine programmer decide which percentages to use? What are >the optimal percentages? [Please assume a mature chess engine (like Crafty), >and not one just started] > >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.