Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Are my tablebases corrupted ???

Author: Eugene Nalimov

Date: 22:54:11 07/11/99

Go up one level in this thread


On July 10, 1999 at 19:41:56, Vincent Lejeune wrote:

>On July 10, 1999 at 19:23:02, Melvin S. Schwartz wrote:
>
>>
>>On July 10, 1999 at 19:01:56, Vincent Lejeune wrote:
>>
>>>On July 10, 1999 at 18:38:28, Sylvain Renard wrote:
>>>
>>>>On July 10, 1999 at 17:34:08, Vincent Lejeune wrote:
>>>>
>>>>>while a looking at this position :
>>>>>
>>>>>8/4KP1k/7r/6R1/8/8/8/8 w - - 0 1
>>>>>
>>>>>
>>>>>Hiarcs 7.32 annouce : Rc6 mate in 6, at the next move : Rh5 Mate in 10
>>>>> and next move : Kg6 mate in 19
>>>>>
>>>>>what's hapenning ???
>>>>>Have someone else the same problem ???
>>>>
>>>>  Hello,
>>>>it seems that the kqrkr files are missing! If you only have the krpkr files
>>>>the program will not promote his pawn!
>>>>  Best regards,
>>>>    Sylvain Renard
>>>
>>>I've verified , you're totaly right.
>>>I warn all the Hiarcs 7.32 user : Hiarcs 7.32 is unable to win KRPvKR : it
>>>annouce the distance to mate but never promote the pawn ...
>>Is this a case where you weren't able to copy all the tablebases from the CD?
>>And that's why you encountered this problem?
>>
>>Regards,
>>Mel
>
>KQRvKR is not on the CD , probably because it require 233 MByte.
>They had to do choice to provide the more usefull tablebase.
>So, if you want to analyse a KRPvKR endgame no problem but if you want to play
>this endgame you must to generate KQRvKR , the tablebase generator (tbgen.exe)
>is on the Hiarcs CD.
>
>End of the story :)
>
>See you soon ...

Yes, but at least they could do the simplest trick: when

(1) Position on the board is TB position and is "Mate in M",
(2) Among the children (moves that can be done from the root position) that are
TB positions there is no "Mate in M-1",
(3) There are children that are captures or promotions and are TB misses (i.e.
there is no corresponding TB),

Then the right way to play is to turn off TB probing for N pieces (where N is #
of pieces OTB) and start normal search.

Amazing that so many good programmers made exactly the same bug. In Bob's case
at least it's easy to explain - he has all the TBs, gives his program for free,
and always can say "download more TBs". But for commercial program that is
unforgivable...

Eugene



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.