Author: David Rasmussen
Date: 04:36:53 12/07/00
Go up one level in this thread
On December 06, 2000 at 12:33:09, Robert Hyatt wrote: >On December 06, 2000 at 08:49:06, David Rasmussen wrote: > >>To compare hashhit success, we have to agree on some way to count this. >> >>In a one probe scheme it is obvious how this is done. But in multiprobe schemes, >>it isn't. For example, in crafty, hashprobes is incremented once for every call >>to probe, but in every call, the hit counter can potentially be incremented >>twice, because it probes twices. This can result in hashhit percentages like >>167% in the endgame. Which is meaningless. > > >This should be very uncommon. Because I take care to not duplicate entries >if possible. There are a couple of ways around the tests, but it doesn't >happen very often. I can't recall seeing a 167% unless you only do a 3 ply >search after having done a 20 ply search... It happens frequently in the endgame with Crafty. Try running a 1 second search in FINE70 and your hashsuccessrate will be about 130%
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.