Djinn 0.979 Available (Win x64, 32 and Linux x64)

[Tom Likens]

It's late and I probably should go to bed, but instead I'm going to pull the trigger on releasing
a new version of Djinn. In my testing it seems about 40 elo better than 0.971, so it should be
the strongest version yet, but since that is an extremely low bar, I think both Houdini and Moron
are safe for the moment.

http://webpages.charter.net/tlikens/index.html

I find other programmer's change logs interesting, so here are the changes since the last release:

Code: Select all

20-JAN-2013 (Ver. 0.979)
 ------------------------
 S  o Changed the history table from "HistoryTable[side][piece][from-to]"
      to "HistoryTable[side][piece][to]".  After extensive testing the
      difference between the two methods was indistinguishable.  But, 
      that's not the end of the story; the second method is far more cache-
      friendly on modern CPUs, so that overall it's a winner.  
 S  o Tweaked the extensions a bit (nothing too major).

 18-JAN-2013
 -----------
 S  o Simplified how killers are handled.  Djinn has always used two 
      killers with a fail-high counter for each move.  After a lot of
      CCC discussion I decided to remove the counter and simply always
      replace the first killer with the new fail-high move while moving
      the previous #1 slot killer into slot #2.  This simplified things
      and in testing works at least as well as the old method (if not
      better).  Anyway, simplier is always better so it's here to stay.
 B  o Minor bug fix in the routine sorting the mundanes.  It was only
      preserving the MVV/LVA order for moves with an equal SEE value.
      This is now fixed so that the MVV/LVA order is preservd across 
      all winning/even mundanes.  

 29-DEC-2012
 -----------
 S  o Modified the way the aspiration window is searched.  The program
      gradually expands the window before jumping to +/-MATE.
 E  o Changed the king safety routines so that they are now checked in
      the endgame as well as the middlegame, if the enemy still has 
      their queen.  Also fixed a minor bug in how the king's pawn cover 
      was being calculated.  The pawn cover for both black and white, 
      (wheter the player castled kingside or queenside) is now scored as 
      if they had castled to the white kingside.  I know this sounds a bit 
      confusing, but it works better and really is more consistent.  It 
      also makes tuning the bonus (or penalty) easier.

 24-NOV-2012
 -----------
 I  o Added the "prevfen" command.  It's analogous to the "nextfen" 
      command to allow a user to walk backwards after loading a EPD file.
      The command can be invoked with "prevfen/prev/p".  I also modified
      the code that loads an EPD file to check that all the FEN strings
      are legal when the file is read and to report any errors it finds.

 21-NOV-2012
 -----------
 E  o Reworked the code that detects a draw when one side only has a 
      rook's pawn and the wrong-colored bishop.

As always please let me know if you find any bugs etc. And of course as always thanks
for putting up with this nonsense.

best,
--tom

[Werner]

Thanks a lot Tom for the release !

I wish you good luck, I think we can start our tests on Weekend.

best wishes
Werner

[ThomasJMiller]

Thanks for your engine. I also would like to thank you for your Computer Chess Books List. I have bought some of them in the last two years (the not too technical ones).

[Tom Likens]

Thanks for your engine. I also would like to thank you for your Computer Chess Books List. I have bought some of them in the last two years (the not too technical ones).

Thanks Thomas,

I really enjoy books and I've been lucky enough to have access to a number of the more
classic titles, so it's my small way of giving back. I do need to update the prices and some of
the links though. A number of the books that were previously hard to find have become available,
(for example, "Computers, Chess, and Cognition" and "Chess Computer Compendium"
are now available in paperback at a more reasonable prices).

I wish they would reissue Heinz's excellent book "Scalable Search in Computer Chess".
It's hard to find and expense when available. I actually gave away an extra copy I had to
someone on this forum (long ago) but I can't remember who was the lucky recipient.

That copy is probably worth about $200 today!

regards,
--tom

[ThomasJMiller]

Thanks for your engine. I also would like to thank you for your Computer Chess Books List. I have bought some of them in the last two years (the not too technical ones).

Thanks Thomas,

I really enjoy books and I've been lucky enough to have access to a number of the more
classic titles, so it's my small way of giving back. I do need to update the prices and some of
the links though. A number of the books that were previously hard to find have become available,
(for example, "Computers, Chess, and Cognition" and "Chess Computer Compendium"
are now available in paperback at a more reasonable prices).

I wish they would reissue Heinz's excellent book "Scalable Search in Computer Chess".
It's hard to find and expense when available. I actually gave away an extra copy I had to
someone on this forum (long ago) but I can't remember who was the lucky recipient.

That copy is probably worth about $200 today!

regards,
--tom

I've looked for Computer, Chess and Cognition for quite some time on Abebooks but I always found used copies for 150$. Nice to know it's availble a reprint now.

[Tom Likens]

I've looked for Computer, Chess and Cognition for quite some time on Abebooks but I always found used copies for 150$. Nice to know it's availble a reprint now.

Here's the Amazon link for the new paperback version:



It's still not free (approx. $79) but still somewhat cheaper than tracking down a clean copy of the original hardback.

[Elorejano]

Hi, Tom!
For some reason, Djinn 979 cant open its own book under winboard, that its the same as always
WindowsXP 32 bits
Thanks for your engine

[Tom Likens]

Hi, Tom!
For some reason, Djinn 979 cant open its own book under winboard, that its the same as always
WindowsXP 32 bits
Thanks for your engine

Hello Carlos,

Thanks for the bug report. I'll look into it first thing tomorrow. I may have a few more
questions after I dig into the problem a bit more.

regards,
--tom

[Ron Murawski]

Thanks for your engine. I also would like to thank you for your Computer Chess Books List. I have bought some of them in the last two years (the not too technical ones).

Thanks Thomas,

I really enjoy books and I've been lucky enough to have access to a number of the more
classic titles, so it's my small way of giving back. I do need to update the prices and some of
the links though. A number of the books that were previously hard to find have become available,
(for example, "Computers, Chess, and Cognition" and "Chess Computer Compendium"
are now available in paperback at a more reasonable prices).

I wish they would reissue Heinz's excellent book "Scalable Search in Computer Chess".
It's hard to find and expense when available. I actually gave away an extra copy I had to
someone on this forum (long ago) but I can't remember who was the lucky recipient.

That copy is probably worth about $200 today!

regards,
--tom

Hi Tom,

That would be me as the recipient. You created a contest: whoever responded to your post first would win the book. I happened to be online and refreshed the page just after you posted. I was very lucky. I refer to Heinz's book several times each year. Thanks again!

Best regards,
Ron

[Tom Likens]

Hi, Tom!
For some reason, Djinn 979 cant open its own book under winboard, that its the same as always
WindowsXP 32 bits
Thanks for your engine

Carlos,

This could be a configuration file issue. One quick question, in the Djinn configuration file do
you have these options set correctly?

Code: Select all

opening-book=[Path to the opening book]/djdjinn.bin
use-opening-book = true

It's an easy mistake to make, I've done it myself recently. Also, make sure you're not inadvertently
using the -xbook option on the command line, (unlikely, but I thought I'd mention it).

--tom