Author: Gian-Carlo Pascutto
Date: 14:18:29 12/02/01
Go up one level in this thread
On December 02, 2001 at 17:03:00, Severi Salminen wrote: >>Two things. First, if your pawn structure evaluation is pretty simple, hashing >>may be a loser for you. Because you update the signature on every pawn move, >>but only evaluate pawns at endpoint positions. If the signature update adds >>up to more "cost" than the evaluation it replaces, you lose some. > >Yep, this seems to be the case: too simple to gain anything from hashing. It >helped a little to use 32-bit key instead of 64-bit. But overall it seems that I >have to wait until my pawn evaluation is more complex. Or then I'm so excellent >programmer that my algorithms are so fast that no need for hashing :) > >I might try to add some pawn island evaluation to see how it affects - it should >narrow the gap. Also I noticed that on my slow (300MHz) computer, even a table >of 65536 entries is enough. If I perform a 11 ply search in position after 1.e4 >e5 2.d4 d5, my pawn hash table still has about 10000 empty slots. Is this too >low? The hit rate was about 96%. After 9 ply search the table was less than >half-full. I allways replace an entry when there is no hit. Two things: a) you can include your kingsafety in the pawnstructure evaluation as well, or at least that part that depends on pawnstructure. You can calculate several kingsafety values for 'if the king would be there' so you do not have to hash the kings position, but just pick out the right one depending on its position b) as you already noticed, there is no sense in making the table big c) you can put loads of stuff in that table, and it's all nearly free to calculate -- GCP
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.