Author: Tom King
Date: 12:57:38 03/15/99
Go up one level in this thread
On March 14, 1999 at 21:26:07, Robert Hyatt wrote: >On March 14, 1999 at 18:02:51, William Bryant wrote: > >>I have a question on move ordering. >> >>The following list is how I think the order should be for determining the moves >>to be searched. I know that as I add PVS, NEGASCOUT, or other variations of >>this theme to my search function, move ordering is important. Any help with >>this list in terms or order of importance or includion/exclusion would be >>appreciated. >> >>1. Hash table move >>2. PV move > >1/2 should be the same move, if you are searching the PV. > > >>3. Killer move/moves >>4. Promotions >>5. Captures > >I don't separate promotions and captures. I lump 'em together, but figure >out which seems to improve the material balance the most. IE I would play >e8=Q if e8 is undefended, rather than a capture that captures less than a >queen. I use my SEE code to sort these moves. > >Also you do _not_ want to do killers before captures. Captures are far more >likely to refute the previous move since an exhaustive search leaves the moving >piece 'hung' many times, or else leaves something else hung. > > Bob's point is important. A long time ago, I did killers before captures, but it was a big loser. Do your captures first! (This is actually mentioned in the famous Chess4.5 paper, written in the 1970's, and still holds today). Rgds, Tom > > >>6. All other moves. >> >>I realize that in some positions, some of these moves may end up being the same. >> I.E. the hash table move and the PV move may be the same move. >> >>Thanks in advance. >> >>William >>wbryant@ix.netcom.com
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.