Author: Robert Hyatt
Date: 20:06:29 03/14/04
Go up one level in this thread
On March 13, 2004 at 16:43:23, F. Huber wrote: >On March 13, 2004 at 14:23:15, Dieter Buerssner wrote: > >>On March 13, 2004 at 11:46:51, F. Huber wrote: >> >>>So what will you be trying next (and the software should take care of)? >>>Adding or removing memory modules, while the computer in ON >>>and Windows is running? ;-) >>>And then writing here again:"Windows bug - system crashes if ..." >> >>I don't see too much similarity in your example and the original problem. I >>think, the software should take care there. The main difference is: the software >>itself allows you to change the hash, so it is under its control (unlike a >>hardware change). Also, for example the UCI protocol explicetly disallows the >>setoption command during search. If you change any parameter the search will be >>stopped, and then new parameters (for example hash size) are sent to the engine. >>In my experience, this works well under Shredder Classic. It allows seemingly to >>change parameters during analysis, but stops the search, before they are really >>changed. >> >>Regards, >>Dieter > >Of course you can do whatever you want when running any software, >but one thing should be clear: >If you do a stupid action, don´t wonder about a stupid REaction! ;-) > >Regards, >Franz. I agree Changing the hash size in the middle of a search goes _beyond_ idiocy. Why? How can it work? Table entries are at the right location. If you change the hash size, suddenly they are at the wrong location as we would have used more bits for the table address. Do you move every entry? Might be impossible. Do you store the entire 64 bit hash signature? I do. Not everybody does. The idea is just ridiculous from the get-go. :)
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.