Computer Chess Club Archives


Search

Terms

Messages

Subject: EGTB problems

Author: Daniel Clausen

Date: 11:56:39 09/12/99


Hi

I'm trying to use Eugenes EGTBs in my own chess engine and downloaded
the file tbgen.zip from Mr. Hyatts site.

I have the following simple test position:
   4k3/8/4K1R1/8/8/8/8/8/ w - -

   WK=E6, WR=G6, BK=E8, white to move

Now this should be mate in 1 in my opinion. :) But when I probe EGTB I
get the value 32767. I tried it with similar positions and always got
the same value.

I'm not sure whether it already fails during index-calculation or while
probing.


Indexing
--------

index = PfnIndCalcFun(iTb, stm) (wpP, bpP, epSquare, fInvert);

When I call this function, the parameters are set as follows:

  iTb      = 4
  stm      = 0
  wpP      = <pointer to squares of white pieces>
  bpP      = <pointer to squares of black pieces>
  epSquare = 127
  fInvert  = 0

I think the wpP and bpP are correct, because the macros produce the following
output:

  SqFindKing(wpP)   = 44
  SqFindKing(bpP)   = 60
  SqFindOne(wpP, 4) = 46

And with these parameters the resulting index I get is 22247.
Can anyone reproduce this index by chance?


Lookup
------

tbScore = L_TbtProbeTable(iTb, stm, index);

  where iTb   = 4
        stm   = 0
        index = 22247

  ==> tbScore = 32767


I'd be happy if anyone could tell me whether the index is already broken or
only the resuling score.

cu,
 -sargon


PS.   I hope I didn't make any silly copy/paste errors...
PPS.  I created the TBs myself and compared them to what's available on
      Mr Hyatts website - and they were identical.
PPPS. I tried to reproduce the above things with Crafty, but unfortunately
      Crafty prefers to hang in InitializeAttackBoards() instead. I'm running
      Linux but on PPC, so.. that could explain this weird behaviour... *shrug*



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.