Author: Tim Foden
Date: 12:01:33 07/12/03
Go up one level in this thread
On July 12, 2003 at 12:27:25, Russell Reagan wrote: >On July 12, 2003 at 03:57:23, Tim Foden wrote: > >>I have been using a dual-table approach >>(primary depth based, secondary always replace)... > >Have you tried node count based instead of depth based? I thought that was shown >to be a better replacement scheme. I haven't exactly heard any results about this, but it seems like it may be a sensible approach to try. I will probably get around the trying it at some time. There are 2 things stopping me... (1) I don't have the node counts handy, although it would be fairly trivial to add, and (2) I don't have space in the hash record to store them (12 bits available), so I'd probably have to go for some log(nodes) scheme. >Also, when using two tables like this, if you store an entry in the depth based >table, do you also store it in the always replace table? This seems like it >would be wasteful as you write to both tables. Nope. I compare with the depth. If it passes, it goes in the primary, and the original primary record is moved to the secondary, if it fails it just goes in the secondary table. I don't have code to restrict the primary and secondary to only have one instance of a particular hash key, as my tests made this look slightly worse. Cheers, Tim.
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.