Author: Robert Hyatt
Date: 18:39:31 01/19/98
Go up one level in this thread
On January 19, 1998 at 19:11:38, Stuart Cracraft wrote: >On January 19, 1998 at 17:37:34, Robert Hyatt wrote: > >>On January 19, 1998 at 17:09:00, Stuart Cracraft wrote: >> >>>What is the typical hash table successful percentage retrieve value? >>> >>>I'm looking for this number for at least one other program for a >>>typical search or group of searches: >>> >>> found-in-hashtable * 100 >>> ------------------------ >>> total attempts to probe hashtable >>> >>>--Stuart >> >> >>Depends on what you are counting, and the position you are searching. >> >>If you are simply counting positions where the hash signature matches >>what is in the table, you should get 25% and higher. If you are >>counting >>positions where the hash table hit actually stops the search early due >>to >>a fail high/low or exact score match, this should be around 10% and get >>better as depth increases. > >I assume your 25% case reflects that the called-depth >= stored-depth >for the store-entry case and stored-depth < called-depth for the >retrieve-entry >case where stored-depth is the depth currently in the hash table and >called-depth is the depth supplied to the store-entry/retrieve-entry >routines. > >--Stuart it applies to *all* probes attempted. If the keys match, they count in the 25%. But not all such positions are useful. Useful hits is in the 10% range for early middlegame...
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.