Author: Robert Hyatt
Date: 08:40:35 10/27/05
Go up one level in this thread
On October 27, 2005 at 10:32:30, Uri Blass wrote: >On October 27, 2005 at 10:10:12, Robert Hyatt wrote: > >>On October 27, 2005 at 04:39:22, Uri Blass wrote: >> >>>suppose that I give you 100 random positions from games. >>> >>>How much do you need to calculate fruit2.1's static evaluation of all positions >>>with no computer help. >>> >>>Uri >> >> >>That's pretty pointless. A bubble-sort is simple compared to a heapsort. Yet >>how long would it take _you_ to bubble-sort 1 million entries? Or even 100? >> >>I could think of other ways to measure "simplicity" (since it is a relative >>term). >> >>lines of code. >> >>number of unique eval terms >> >>number of instructions executed to do a single evaluation. >> >>etc... > >I understand your point. > >Note that it is hard to use lines of code because of the following problems. > >1)There are ASSERT in the code and without them the evaluation could be the same >with less lines. I would not count asserts, nor comments, nor declarations of data values, since int x; int y; int z; would therefore look more complex than int x,y,z; However, in software engineering, lines of code has always been considered a bad measure of anything, since the _same_ algorithm can be expressed many different ways that have differing numbers of lines of code, while doing _exactly_ the same thing... > >2)There are empty lines in the code. > >3)The evaluation code is not in a single file and I can see evaluation code in >material.cpp,eval.cpp,pawn.cpp,recog.cpp pst.cpp and maybe more code. > >these files include often the same lines for example >#include "piece.h" and it is not clear if to count this line once or more than >once. > >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.