Author: Robert Hyatt
Date: 05:50:05 06/28/01
Go up one level in this thread
On June 27, 2001 at 13:38:45, Vincent Diepeveen wrote: >On June 27, 2001 at 00:02:09, Robert Hyatt wrote: > >>On June 26, 2001 at 12:00:26, Vincent Diepeveen wrote: >> >>>On June 26, 2001 at 11:06:35, Dan Andersson wrote: >>> >>>that's why bitboards get still used by people at 32 bits processors. >>>They never took effort to measure how fast/slow it is. >> >> >> >>Or maybe we just look to the future. The world will be 64 bits in another >>year or two. Why stay in the stone age? > >>People were saying the same thing about 32 bit programs 10 years ago... > >I port DIEP within 1 week of work to 64 bits if everyone on this planet >has a 64 bits machine. > So your scores will be 64 bits wide? your chess board values will be 64 bits rather than 4 bits wide? All your "patterns" will use 64 bit data values rather than 32 or 8 bit ones? We are talking about two _different_ things. It does no good to use a 64 bit machine and pump around 1-byte board values. >> >>> >>>>In most cases I get all the information I need from one or a few indirect memory >>>>lookups, that's not too slow IMO. In my case, the Attackboard is used in >>>>conjunction with an efficient conventional representation. I do feel that it's >>>>correct to incrementally update variables that are used to either improve move >>>>ordering or create cutoffs. As for speed of implementation I cannot really >>>>compare figures, since I reuse a lot of data i.e. the behaviour will be >>>>asymptotic (and very fast) and not representative of its behaviour on a >>>>completely new position. But on closely similar positions (for example a >>>>sequence of positions in a line searched) it has a good performance. >>>> >>>>Regards Dan Andersson
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.