Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Stockfish 1.6s -Dann Corbit's masterpiece
- - By Silvian (***) Date 2009-12-29 08:13 Edited 2009-12-29 08:15
Take a look to this brother ( :):):) ) of Rybka 5/6:

http://cap.connx.com/chess-engines/new-approach/Stockfish.exe.bz2

Silvian
Parent - - By Permanent Brain (*****) Date 2009-12-29 08:31 Edited 2009-12-29 09:08
Seems to be 64 bit only (?). The related CCC postings mention a kind of improved nullmove (I don't understand much of it).

Edit: There is a 32 bit version too which I found in:

http://talkchess.com/forum/viewtopic.php?t=31330

It's called Stockfish 1.6s (Stockfish32.exe). A quick comparison from the starting position; pls note that it's from an old CPU:

Analysis by Stockfish 1.6 JA / D945 (~dual P4), 2*3.4 GHz, 256 MB hash:

1.Nf3
  +/-  (0.72)   Depth: 1   00:00:00
1.Nf3 Nf6
  =  (0.12)   Depth: 2   00:00:00
(...)
1.e4 Nf6 2.e5 Nd5 3.Nf3 d6 4.d4 dxe5 5.Nxe5 Nd7 6.Nxf7 Kxf7 7.Qh5+ g6 8.Qxd5+ e6 9.Qg5 Nf6
  +/=  (0.64)   Depth: 15   00:00:08  5792kN
1.e4 Nc6 2.d4 d5 3.exd5 Qxd5 4.Nf3 Bg4 5.Nc3 Qa5 6.Be3 0-0-0 7.h3 Bxf3 8.gxf3 Nf6 9.Bc4
  +/=  (0.44)   Depth: 16   00:00:16  13580kN
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.c3 Nf6 5.d4 exd4 6.e5 Ng4 7.cxd4 d5 8.exd6 Bxd6 9.Bg5 f6 10.Qe2+ Qe7
  +/=  (0.32)   Depth: 17   00:00:32  26850kN
1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nf3 Nxe4 5.d3 Nf6 6.Nc3 Nc6 7.Be2 Be7 8.0-0 0-0 9.Re1 Bf5 10.Be3
  =  (0.20)   Depth: 18   00:00:58  47533kN --> ~820 kN/s

Analysis by Stockfish 1.6s (after closing and restarting the interface):

1.Nf3
  +/-  (0.72)   Depth: 1   00:00:00
1.Nf3 Nf6
  =  (0.12)   Depth: 2   00:00:00
(...)
1.e4 e5 2.Nf3 Nf6 3.Nxe5 Qe7 4.d4 d6 5.Nf3 Qxe4+ 6.Be3 Ng4 7.Nc3 Nxe3 8.Nxe4 Nxd1 9.Rxd1 Bf5 10.Bb5+ Nc6 11.Bxc6+ bxc6
  =  (0.20)   Depth: 15   00:00:04  3466kN
1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nf3 Nxe4 5.d3 Nf6 6.Be2 Nc6 7.0-0 d5 8.Nc3 Bd6 9.Nb5 0-0 10.Nxd6 Qxd6 11.Be3
  +/=  (0.32)   Depth: 16   00:00:08  6923kN
1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Bb5 Bb4 5.Bxc6 dxc6 6.Nxe5 Qd6 7.d4 c5 8.Be3 Nxe4 9.0-0 Nxc3 10.bxc3 Bxc3
  +/=  (0.28)   Depth: 17   00:00:12  10383kN
1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Bb5 Bb4 5.0-0 0-0 6.d3 d6 7.Bg5 Be6 8.Nd5 Bxd5 9.exd5 Nd4 10.Nxd4 exd4
  +/=  (0.32)   Depth: 18   00:00:19  17381kN
1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Bb5 Bb4 5.0-0 0-0 6.d3 d6 7.Bg5 Bxc3 8.bxc3 Bd7 9.Rb1 h6 10.Bxf6 Qxf6 11.Bc4 b6 12.d4
  =  (0.24)   Depth: 19   00:00:28  25143kN
1.e4 Nc6 2.d4 Nf6 3.d5 Ne5 4.Nf3 d6 5.Nxe5 dxe5 6.Nc3 c6 7.Bd3 cxd5 8.exd5
  +/=  (0.44)   Depth: 20   00:00:48  42689kN --> 889 kN/s

This is a huge difference! The node rate is very similiar, but the depth - in this particular test - is +2 plies within a minute each.
Parent - By nebulus (****) Date 2009-12-29 08:59
Looks like the 32-bit version is here.
Parent - - By Gaмßito (****) Date 2009-12-29 12:35
Seems that after the emergence of the great RB engine, other engines are having and will have too a BIG JUMP in strength.

With regard to this ''miracolous'' improvements, everyone here can talk or even share links to download, as if everything were the most normal things in the world?

That's is simply unfair, in all senses, because you know as many other people too, from where comes all these wonderful ideas and impressive improvements. :-)

Why punish the clone, (that has many different things that do NOT have the original R3) but not the programs that also use many ideas of the clone? What an unfair and rare moderator's policy exists here, does not believe you?

Regards,
Gaмßito.
Parent - - By Uri Blass (*****) Date 2009-12-29 12:41
I think that you are clearly unfair to stockfish.

Stockfish is simply improving and
I saw no proof that the improvement came from some engine that his authors are not known.

You claim that many ideas of stockfish came from the clone and you gave no evidence for it.

Uri
Parent - Date 2009-12-29 12:51
Parent - - By Uly (Gold) Date 2009-12-29 12:55
Uri Blass is right, the improvement is due to a new implementation of an old idea and not due to the clone, you can read more about it here:

http://www.talkchess.com/forum/viewtopic.php?t=31329&start=0

The changes between 1.6 and 1.6s are there in plain sight so you can examine them.
Parent - - By turbojuice1122 (Gold) Date 2009-12-29 22:28
True, but as is noted further down, this new implementation was used most recently in Robo. :-)
Parent - - By Uly (Gold) Date 2009-12-29 22:42
Dann's comment on the comment you mention:

"He seems to have gotten away with a lot less math. Maybe it is a better idea.

However, this approach does not account for existing depth and the reductions will be titanic for huge scores (which is perhaps a good thing)."

So it seems Dann thought about using the idea by himself ;)
Parent - - By dcorbit (***) Date 2009-12-29 22:49
I have a friend who can declare that I showed him this idea in 1997.
Parent - - By Uly (Gold) Date 2009-12-29 22:57
Yes, I still think it's sad that people see big improvements and assume it was just copy pasting from the clone :(
Parent - By Geomusic (*****) Date 2010-03-11 09:54 Edited 2010-03-11 09:57
Vytron, it's completely justifiable to the public at large considering... People assume the worst upon any success now.

Just my 2 pence..
Parent - By ChessMate (***) Date 2009-12-29 13:15 Edited 2009-12-29 13:18

> With regard to this ''miracolous'' improvements, everyone here can talk or even share links to download, as if everything were the most normal things in the world?
>
> That's is simply unfair, in all senses, because you know as many other people too, from where comes all these wonderful ideas and impressive improvements. :-)
>
> Why punish the clone, (that has many different things that do NOT have the original R3) but not the programs that also use many ideas of the clone? What an unfair and rare moderator's policy exists here, does not believe you?


Are u serious ?

I think u  have not read Dann's statement.
Parent - - By Aliven (***) Date 2009-12-29 08:50
Can we have more information please? Is this an improved Stockfish 1.6?
Parent - - By Uly (Gold) Date 2009-12-29 10:38
Yes, it's meant to improve the scaling, I believe.
Parent - - By Not the real Kasparov (***) Date 2009-12-30 08:26
Guys i have downloaded the apparent correct version of stockfish 1.6s, and it appears to be weaker than the normal stockfish 1.6, has anyone else had these results? I only ran a small amount of games but it's results were worse against normal stockfish and Rybka 3.
Parent - By nebulus (****) Date 2009-12-30 11:36
I had the same results.
1' 1'' 256Mb, stockfish16s-dc vs stockfish16-ja 9-11
1' 1'' 64Mb, stockfish16s-dc vs stockfish16-ja 10,5-17,5
5' 0'' 256Mb, stockfish16s-dc vs stockfish16-ja 8-12
5' 0'' 256Mb, stockfish16s-dc vs stockfish16-ja 5,5-8,5
Though when played in a 4 round 5' 0'' tournament with Rybka 3, stockfish16s-dc scored 3,5/4 stockfish16-ja and 2-2 against rybka.
Parent - - By Uly (Gold) Date 2009-12-30 15:42
Did you download the 32bit version from http://cap.connx.com/chess-engines/new-approach/Stockfish.exe.bz2 ? Because Dann changed the link and now that leads to the new experimental version that may be weaker, try this one:

http://rybkaforum.net/mwf/rybkaattach/46/214046/dannstfsh.zip

The 32bit version from that file.

For the 64bit try here:

http://www.mediafire.com/?1njjnmee22k

And yes, the bad versions are identical to the good versions other than the file's date (but newer or older date doesn't mean anything.)
Parent - By Not the real Kasparov (***) Date 2009-12-31 00:25
Thanks for the download link, this stockfish appears to be stronger fantastic!
Up Topic The Rybka Lounge / Computer Chess / Stockfish 1.6s -Dann Corbit's masterpiece

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill