Author: Tom Likens
Date: 19:08:09 09/18/03
Go up one level in this thread
On September 18, 2003 at 19:58:34, macaroni wrote: >Hi, thanks so much for all your excellent posts to my question. > >I noticed that one of you (Uri perhaps?) said that having a better eval didn't >neccesarily slow down the search too significantly, so what is the fastest way >to test mobility (the thing which seems to slow me down most), at the moment, my >good eval tests it by simply calling move gen for both sides, and counting the >number of moves. This is not that efficient it seems (adds about 17 seconds to a >1,000,000 node search, and doesn't seem to cause significantly more cut-offs). >Is there a better way to do this? perhaps store the number of moves available in >the previous two plys for an estimate? If I take out the mobility code I have, >it speeds up to a reasonable level! > >thanks again! It's tough to speculate without knowing a bit more about your engine's architecture. Are you 0x88 based, bitboard based etc. The most efficient way to accomplish this will depend heavily on the data structures you are using and have available. --tom
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.