Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hash codes

Author: Uri Blass

Date: 03:44:44 11/14/01

Go up one level in this thread


On November 14, 2001 at 06:11:04, Gian-Carlo Pascutto wrote:

>On November 14, 2001 at 05:11:34, Rémi Coulom wrote:
>
>>It happened to me. I had been using only 32 bit (or maybe even a bit more, I do
>>not remember) for years and never noticed any problem. TCB once crashed very
>>mysteriously and I managed to find that this was because of a hash collision. I
>>now use more bits and have made my hash algorithm a bit more collision-proof.
>
>So, basically, it took a long time before it got you into
>trouble, and even then it was because you were not carefull
>to account for the possibility of a collision?
>
>(I'm very tempted to run my own tests now)

The interesting question is if bigger hash tables are better or worse in finding
moves(it is possible to take a lot of positions from games and give 2 versions
of the same program to search for some minutes every position and compare the
time that the program need to get the same move in cases that they converge to
the same move(cases that they do not vonverge to the same move need to be
checked)

The demage can be also finding the right move slightly slower and not only not
finding the right move so the fact that someone did not discover problems does
not mean thet there were no practical problems.

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.