Author: Miguel A. Ballicora
Date: 09:19:41 11/04/02
Go up one level in this thread
On November 04, 2002 at 10:28:20, Tom Likens wrote: >On November 04, 2002 at 04:32:20, Severi Salminen wrote: > >>>I can see randomize when I click F1 only as a visual basic command and I work in >>>C. >>> >>>I tried srand(0) before filling the array but it does not help and >>>zobrist[0][0][0] is different in debug mode and in release mode. >> >>Are you sure? I think rand() should return the exact same values if you use the >>same seed (srand()). Are you sure you run srand(0) and fill the table only once >>in both versions? There seems to be a bug somewhere. >> >>Severi > >About 5 years ago I got tired of my program producing different >results when run on different architectures due to the various >implementations of the random number generators. So instead of >creating the random numbers on the fly I wrote a program to >produce the zobrist values and stored them in a file that is >read once upon initialization. Excellent idea. Either that, or an array that is defined, or to have your own random generator that does not depend on anything from the system, but the seed that you provide. That is extremely important, othewise, even the opening book can change if you change compilers! Miguel >The nice thing about this is that you can go crazy, if you like, >ensuring that these numbers have all the nice properties that >random numbers should have (i.e. hamming distance etc.). The >really nice thing is that now the random values are predictable >no matter which machine or operating system or compiler your >engine is run under (of course whenever the hash table is >involved *nothing* is predictable, but that's another story). > >regards, >--tom
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.