Author: Dan Andersson
Date: 12:40:40 09/11/04
Go up one level in this thread
On September 11, 2004 at 14:49:41, Stuart Cracraft wrote: >On September 11, 2004 at 12:13:36, Sune Fischer wrote: > >>On September 11, 2004 at 00:08:20, Stuart Cracraft wrote: >> >>>Hi, >>> >>>I added keeping a triangular pv in main search and quiescence >>>to compare it with the output of my walk-the-hashtable-pv. >>> >>>The two differ frequently but quite often are also mostly >>>identical all the way through. >> >>Don't forget to check the hash flag, that the moves are actually PV moves. >> >>Mostly you get them overwritten with upper or lower moves, those should not go >>in the PV. >> >>>Which should I trust? Seems like the hash table is getting >>>overwritten with other variations (not sure why). What >>>kind of scenario would cause that? My algorithm is >>>length >= depth to replace. >> >>That's not a very good replacement scheme, if you only have a single bin I'd >>recommend using replace always. > >This gave a nice improvement on a Thinkpad laptop of 237 solved of WAC >@ 1 second per, to 244. > >I guess recency is more important than depth!!! I don't know why I never >even considered to replace always. Didn't even test it. I always had >assumed that depth was more important than recency. Bad assumption A quick question. When you write recency do you mean temporal locality? And when you think about it it does make a lot of sense to always replace since the temporal locality fitness of the hash is higher whith that scheme. > >Nowadays, do most use 2-tier or ? If so or whatever, what are the preferred >replacement algorithms? A two tier scheme of depth/always replace is in general hard to beat. The lessened capacity of the more local always replace is more than compensated by the work saved by the depth replace hash. The two hashes do not necessarily have to be of equal size. MvH Dan Andersson > >Thanks, >Stuart
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.