Author: Robert Hyatt
Date: 17:20:25 02/20/05
Go up one level in this thread
On February 20, 2005 at 13:52:31, Uri Blass wrote: >On February 19, 2005 at 23:16:41, Robert Hyatt wrote: > >>On February 19, 2005 at 20:51:15, Peter Skinner wrote: >> >>>On February 19, 2005 at 20:26:02, Robert Hyatt wrote: >>> >>>>Note there are "6 man egtbs" and there are "new 6 man egtbs". The new ones >>>>require an updated egtb.cpp, which is not going to be a part of any engine that >>>>is over a month old, as that is when the new egtb.cpp was released... >>> >>>This brings up a good question. Is it safe to mix old and new 6 man egtbs >>>together? >>> >>>Or will they all be converted to the new format? >>> >>>Peter >> >> >>Format isn't new. Just the old EGTB probe code has some egtbs flagged as 8 bits >>when they have to be 16 bits. The format for all files is exactly the same >>except for this, and we've had 16 bit tables already. Any that have a mate in >>over 125 or 126 (not sure which) require 16 bits for the scores. > >Why 16 bits? > >I think that if there is no mate in 255 or longer mate >9 bits can be enough > >I also do not understand why mate in 126 means that 8 bits are not enough You can store +127 to -128 in 8 bits. What the missing values are used for I don't remember as I haven't looked at the code in a long time. Who would want to index into 9 bit entries? A bit of a headache. :) > >mate in 1,....mate in 126(126 scores) >loss in 0,loss in 1,...loss in 126(127 scores) >draw(1 score) > >number of possible scores is 254 that is smaller than 256 so it seems that 8 >bits are enough. Don't know what you mean. Scores beyond mate/mated in 127 are plentiful in some of the new endings. > > >Uri
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.