Author: Robert Hyatt
Date: 08:36:59 10/22/02
Go up one level in this thread
On October 22, 2002 at 10:13:11, Sune Fischer wrote:
>I'm trying to implement support for the
>Nalimov endgame tables and are having a minor(?) problem.
>
>By reading the source of Beowulf and the probe.txt file
>I have managed to initialize the tables and probing is also
>partly successful. It findes the table, but probing it
>doesn't work:
>
>value = L_TbtProbeTable(iTB, colour, index);
>if(value == L_bev_broken) return 0;
>
>I get the return here.
>
>Since the whole setup is more or less a black box for me,
>I have no idea what can be the cause of this, any suggestions?
>I have also a few questions, first of which is Beowulf specific.
I would suspect that you are not passing the position in correctly. And what
you
pass in might well be a "broken" position (two pieces on the same square, no
pieces
on any square, etc...)
>
>These declarations cause two compiler errors in my code:
>
>int ProbeEGTB(Board *B, int *score, int ply) {
> int *wp, *bp; // 1st and 2nd argument to PfnIndCalcFun()
>...
>}
>
>The compiler informs me that it expects unsigned int*. That seems to work, but
>how can Beowulf use a wrong data type and still compile? I think maybe it is
>related to the L_bev_broken error I get.
>
>Also why is egtb.c 335 kB in Beowulf and only 156 kB in Crafty and where are the
>macros of INDEX and square? Have the egtb.c files been modified?
>
>When I use the default tbindex.cpp it doesn't work without me manually
>declaring some macros. Ok, pretty messy, but what is the "right" thing to do?
>
>Finally, if I want to support 6 man tables, do I need to do anything else than
>#define C_PIECES 4 instead of 3 (and of course increase size of WhitePieces and
>BlackPieces arrays to hold them all)?
>
>-S.
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.