Author: Severi Salminen
Date: 10:06:02 03/12/01
Go up one level in this thread
>Thanks for your reply. After taking your advice, >I saw that the executable file was much larger. >In a singe game, the new version lost, but >this may very well be bad luck, so I think I will >play a match of, say, 20 games, to compare >the same program compiled with two different >settings. Not a good idea - you will get no idea of which one is faster or stronger. Instead make your program search a certain position to a fixed depth and compare NPSs of different version. This is a "good" way to use NPS as comparison and this is a good way to use single position for comparison. This way you can also check that the new version is behaving properly by checking that the node count matches for both versions (for example if you try out a new assembly routine for move generator etc.). Severi
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.