Stockfish 2.0 Available

[kingliveson]

Stockfish 2.0 JA (UCI) by The Stockfish Team.

This is somewhat unexpected !


Stockfish 2.0 is available from Jim's site:

http://homepages.tesco.net/henry.ablett/jims.html

Diff stats from 1.9.1 says:

54 files changed, 2130 insertions(+), 2812 deletions(-)

So almost 700 lines of code removed for this release.


Actually we didn't plan to release now, but it happens that our test
framework will be down for a couple of weeks due to new year holidays
and so, instead of just waiting two weeks, we thought it could be more
useful to release.

The release number 2.0 is to be intended as the new year release not
as a much improved version from 1.9.1: we expect some ELO gain but
nothing earth shattering Wink

Among the new user-visible changes we would like to note:

- Maximum allowed threads increased to 16 (it was 8)

- Added new option "Use Sleeping Threads" that modifies how threads
are managed during multi CPU searches. Setting to ON should greatly
help on Hyper Thread enabled machines and _could_ help also for other
SMP cases. It is off by default because is not tested a lot (also
because we don't have big hardware) and results depend on the specific
machine. So to be on the safe side we release 2.0 with the option
disabled (here specific tests by users would be _very_ welcomed and
perhaps there could be some nice surprise especially on powerful
hardware).

- Fixed some bugs in Multi PV (courtesy of Peter Petrov), hopefully
should work now, but please report any bug you may find in this area.


On the technical side :

- We now use RKISS random generator for hash keys (courtesy of Heinz
van Saanen) instead of mersenne, it is as good as the old one and is
much smaller and simpler.

- For people compiling themselves, stockfish 2.0 signature is 7224363.
It means that running from command line 'stockfish bench' you should
get at the end 7224363 searched nodes. If this not happens please
report back, it means SF is miscompiled.


As usual a big thanks to Jim Ablett for his wonderful compiles and a
special thank to the people that allow us to use a powerful testing
framework that has really helped us to speed up development and to
achieve a better code quality overall.

Have fun and happy new year 2011 !
SF Team

[kingliveson]

I had a short encounter with Tord on chessbomb during recent London Chess Classic tournament and was a little disappointed when he responded to my query about EGTB support that there were no current plans. He did say more than likely it will be bitbase though.

[Howard E]

Great news. Thanks to the Stockfish team.
I rename the engine on my computer to "Infinito" with a cool
infinity logo. Just as Stockfish has common grounds in Finland and Italy, so too does Infinito
In fin(Finland) ito (Italy) ... I know it's a stretch but I think it plays 10 elo points better with the name change

[stvs]

stockfich 2 seems is weaker than 1.9, 1.8 ??!! according ippo forum tests ...strange
meanwhile in rybkaforum mods allowed one guy to accuse stockfish as a clone engine !!!!
that guy claimed that only chessbase products hiars,fritz,shredder,and rybka are original !!!
chessbase campain vs free engine under way ?

http://rybkaforum.net/cgi-bin/rybkaforu ... ?tid=20098

from "chess" :

"When I was coding my first chess program, I had to work out a lot of stuff for myself. These people just have to copy and paste for hacking together some new engine. Frans Morsch, Stefan Meyer-Kahlen, Mark Uniacke, Vasik Rajlich, those are the guys I call chess programmers."

[robbolito]

Engine Score Ho St S-B

Code: Select all

1: Houdini 1.5.x64  30.0/40 ······· 25-5-10  300.00
2: Stockfish 2.0.64 10.0/40 5-25-10 ·······  300.00

40 games played / Tournament is finished
Name of the tournament: Arena tournament
Site/ Country: i7 870 @ 2.93 GHz, Uruguay
Level: Blitz 10/0
Hardware: Intel(R) Core(TM) i7 CPU Q 720 @ 1.60GHz 1600 MHz with 4.0 GB Memory
Operating system: Windows 7 Home Premium Home Edition (Build 7600) 64 bit

[notyetagm]

Engine Score Ho St S-B

Code: Select all

1: Houdini 1.5.x64  30.0/40 ······· 25-5-10  300.00
2: Stockfish 2.0.64 10.0/40 5-25-10 ·······  300.00

40 games played / Tournament is finished
Name of the tournament: Arena tournament
Site/ Country: i7 870 @ 2.93 GHz, Uruguay
Level: Blitz 10/0
Hardware: Intel(R) Core(TM) i7 CPU Q 720 @ 1.60GHz 1600 MHz with 4.0 GB Memory
Operating system: Windows 7 Home Premium Home Edition (Build 7600) 64 bit

Goddamn Houdini 1.5 is strong.

[Peter C]

I had a short encounter with Tord on chessbomb during recent London Chess Classic tournament and was a little disappointed when he responded to my query about EGTB support that there were no current plans. He did say more than likely it will be bitbase though.

Have fun.

Peter

[Tord]

I had a short encounter with Tord on chessbomb during recent London Chess Classic tournament and was a little disappointed when he responded to my query about EGTB support that there were no current plans.

As far as I know, nobody has yet found a way to use EGTBs effectively. It seems to me that they are rapidly becoming obsolete technology: With CPUs getting faster and faster while hard disks staying almost as slow as before, it's getting more and more difficult to find any useful ways to use EGTBs. Adding EGTB support to Stockfish now would add considerably to the complexity of the program while leaving the playing strength unchanged or even making the program slightly weaker. We don't want to make the program significantly more complex unless we can hope to gain Elo points from doing so.

For those who think that even though EGTBs are not useful when playing games, they would improve the quality of analysis: No, they wouldn't, in general, not even in the late endgame. They would help in some positions, and they hurt in others. If you're analyzing some position where you are sure EGTBs would help, there are plenty of other programs to choose from.

Finally, because Stockfish is open source, anyone who wants is free to add EGTB support himself. As you can see in another thread in this sub-forum, this has already happened with SF 2.0.

He did say more than likely it will be bitbase though.

That's not quite what I said, unless I mistyped. I said that bitbase support was more likely (i.e. more likely than EGTB support), not that it was more than likely. We don't have any very concrete plans about bitbase support either, but bitbases are lot more interesting to us, because it may very well be possible to use them to make the program slightly stronger.