Author: Shep
Date: 03:24:23 07/07/99
Go up one level in this thread
On July 06, 1999 at 11:28:18, Christopher A. Morgan wrote:
>Hiarcs retains hashtable info between moves for future reference.
>In Engine v. Engine in Fritx 5.32 (probably also in Hiarcs 7.32) I assume
>after Hirac's move RAM (hashtable)is cleaned by Fritz for its move. In playing
>around with various hashtable settings and time controls its fairly easy
>to get Fritz to fill hashtables frequently (I have 64MB Ram on Pent II @ 266).
>When that happens Fritz (apparently) goes to Win 98 dynamic swap file
>on hard drive as hard drive lights up and makes "writing, accessing" noises.
>After Fritz moves, Hiarc never accesses hard drive for its move. This
>leads me to believe that Hiarcs "saved" hashtable on its previous move has
>been deleted from RAM, and was never stored on the hard drive, so Hiarcs must
>start from scratch. Right? (for my tests I use same hashtable size for both
An easy way to find out if Hiarcs' hash table was cleared is the following:
play an engine vs. engine match on a single CPU (thus PB disabled) and watch
Hiarcs' depth counter when Fritz made a move which Hiarcs predicted correctly.
If it shows something odd like "10/3" (i.e. full depth greater than selective
depth), Hiarcs is accessing its saved hash table.
If Hiarcs seems to count depth normally ("8/8 -> 8/21 -> 9/9 ..."), it does not
use any saved information.
And I think it has been mentioned on this board that Fritz clears its opponents'
hash table every move, so I suppose there's no way to stop him from doing so.
At least not in automatic eng-eng games.
---
Shep
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.