Author: Russell Reagan
Date: 02:24:54 12/27/02
Go up one level in this thread
On December 27, 2002 at 04:54:42, Dan Andersson wrote: >What Bruce is refering to is the idea that if a hash collision happens you >simply put the offending hash in the next available slot. The possible problem >is that you get a chaining. I.e. a long sequence of occupied slots. But that is >rare. I prefer using a recoding algorithm that computes a new index, the net >result is the same. The hash should be at least twice as large as the depth of >search to get good behaviour. A mean number of probes less than two. One other >idea is to create a new hash table each time a capture is made. The question is >if the extra bookkeeping will make it a win, I'd guess no. Ok. Makes sense so far. So, when you probe, you start with the "normal" index that you get from the hash key (key % size, or whatever), and then compare hash keys until you find a match, or until you hit an entry where the "index" portion of the keys no longer match? Or do you stop probing at a certain number? For example, Vincent says he uses 8 probes. Does that mean he starts at the "normal" index, and iterates until he either gets a hit, or until he has tried 8 slots? Russell
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.