Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Stockfish 9
- - By cghori (**) [us] Date 2017-08-28 00:27
Anyone have any idea when SF9 will be released?
Parent - - By leavenfish (***) [us] Date 2017-08-28 01:23
Sure...just as TCEC starts. :cool:
Parent - - By cghori (**) [us] Date 2017-08-28 03:49
When does TCEC start?
Parent - By Venator (Silver) [nl] Date 2017-08-28 14:36
End of September / beginning of October.
Parent - By Ray (****) Date 2017-08-28 06:48
Or after it finishes ?
Parent - - By Fulcrum2000 (****) [nl] Date 2017-08-28 19:23
As always when the new version is >50 ELO compared to the previous version.
Current dev is 'only' +30 yet, so no release soon.
Parent - - By Geomusic (*****) Date 2017-08-31 15:28
50+ ELO seems like a stretch...do you think it will make it?
Parent - By Fulcrum2000 (****) [nl] Date 2017-08-31 18:21
Sure it will, just not sure if it will be at that point when TCEC ends.
Parent - - By Peter Grayson (****) [us] Date 2017-08-31 22:51

> As always when the new version is >50 ELO compared to the previous version.
> Current dev is 'only' +30 yet, so no release soon.


At which time control, match conditions and hardware base as a standard? Is there a standard for measurement to determine when 50 Elo is achieved? I don't know of one.

PeterG
Parent - - By Fulcrum2000 (****) [nl] Date 2017-09-01 22:45 Edited 2017-09-01 22:49
40000 1m + 0.6s games on the Stockfish test framework
http://tests.stockfishchess.org/tests/view/59a1a55a0ebc5916ff64aba4
Parent - By Peter Grayson (****) [us] Date 2017-09-01 23:34
OK, thanks
Parent - By RFK (Gold) Date 2017-09-05 03:38
I just use the latest version anyhow -never really bother with running 8 as a standalone. Sometimes I run across issues with the most recent version posted but for the most part they seem to do the job.
Parent - - By Fulcrum2000 (****) [gb] Date 2018-01-31 12:23
Probably tomorrow, waiting on the final executables to be generated.
Parent - - By RFK (Gold) Date 2018-01-31 19:31
Author: mstembera
Date: Wed Jan 31 11:41:09 2018 +0100
Timestamp: 1517395269

Stockfish 9

Official release version of Stockfish 9

Bench 5023629
Parent - - By Fulcrum2000 (****) [nl] Date 2018-01-31 20:52
That are not the final executables, those will be released tomorrow via https://stockfishchess.org/

(the new exe's will be functionally identical, but faster)
Parent - By RFK (Gold) Date 2018-01-31 21:14

> the new exe's will be functionally identical, but faster


Fundamentally it is Stockfish 9!  mstembera -stated clearly "Official release version of Stockfish 9 "

I would agree that he has perhaps jumped the gun!
Parent - - By pokerpawn (***) [be] Date 2018-02-01 19:27
I still see SF8 https://stockfishchess.org/
or is it just me? :)
Parent - - By Hurnavich (Bronze) Date 2018-02-01 20:49
Hi, Just refresh the page F5 or CTRL and F5
Parent - By pokerpawn (***) [be] Date 2018-02-02 03:08
thx ,yes i tried refresh , didn't work at first , later it did , weird :-)
- - By Kajordzak (**) Date 2018-02-01 23:25 Edited 2018-02-01 23:27 Upvotes 1
Parent - - By leavenfish (***) [us] Date 2018-02-06 03:43
Anyone know of a good, faster compile yet for SF9?
Parent - - By QueensideCastler (**) [no] Date 2018-02-06 09:56 Upvotes 1
I recommend Cfish 9 by Ronald De Man.

https://github.com/syzygy1/Cfish/releases
Parent - By leavenfish (***) [us] Date 2018-02-07 04:50
Why do you say that? Posts on Talkchess is indicating it is actually slower than SF9.
Parent - - By Peter Grayson (****) [gb] Date 2018-02-07 16:17
The Brainfish compile is the same code as SF9 with the benefit of LP hash. It only becomes Brainfish with the use of the Cerebellum book.

http://zipproth.com/index.html#ui-tabs-1

http://zipproth.com/Brainfish/BrainFish_180201.zip

In single thread mode, it gives exactly the same output as SF9 but significantly quicker dependant upon you hardware and O/S set up.

Stockfish 9 output from start position Q9550 at 3.53 GHz + 1Gb,  single thread

1.d4 Nf6 2.Nf3 d5 3.c4 e6 4.Nc3 Nbd7 5.Bg5 h6 6.Bh4 Bb4 7.e3 c5 8.Be2 g5 9.Bg3 Ne4 10.Rc1 Qa5 11.0-0 Bxc3 12.bxc3 Nxc3 13.Qe1 Nxe2+ 14.Qxe2 0-0 15.cxd5 exd5 16.dxc5
  +/=  (0.36)   Depth: 27/33   00:01:05  66575kN
1.d4 Nf6 2.Nf3 d5 3.c4 e6 4.Nc3 Nbd7 5.cxd5 exd5 6.Bf4 c6 7.e3 Nh5 8.Be5 Nxe5 9.dxe5 g6 10.Be2 Bg7 11.h3 f5 12.Nd4 Bxe5 13.Bxh5 0-0 14.0-0 gxh5 15.Qxh5
  +/=  (0.33)   Depth: 28/35   00:01:20  80780kN
1.d4
  +/=  (0.41 ++)   Depth: 29/35   00:01:26  87148kN
1.d4 Nf6
  =  (0.26 --)   Depth: 29/38   00:02:14  136MN
1.d4 Nf6 2.Nf3 e6 3.c4 b6 4.Nc3 Bb4 5.Bg5 h6 6.Bh4 Bb7 7.e3 g5 8.Bg3 Ne4 9.Qc2 f5 10.Bd3 Bxc3+ 11.bxc3 0-0 12.0-0 c5 13.Nd2 Nxg3 14.fxg3 Nc6 15.e4 cxd4 16.exf5 dxc3 17.Qxc3 Qf6 18.Qxf6 Rxf6 19.fxe6
  +/=  (0.35)   Depth: 29/41   00:02:36  157MN

Brainfish "9" (010218) output from start position Q9550 at 3.53 GHz + 1Gb,  single thread

1.d4 Nf6 2.Nf3 d5 3.c4 e6 4.Nc3 Nbd7 5.Bg5 h6 6.Bh4 Bb4 7.e3 c5 8.Be2 g5 9.Bg3 Ne4 10.Rc1 Qa5 11.0-0 Bxc3 12.bxc3 Nxc3 13.Qe1 Nxe2+ 14.Qxe2 0-0 15.cxd5 exd5 16.dxc5
  +/=  (0.36)   Depth: 27/33   00:00:54  66575kN
1.d4 Nf6 2.Nf3 d5 3.c4 e6 4.Nc3 Nbd7 5.cxd5 exd5 6.Bf4 c6 7.e3 Nh5 8.Be5 Nxe5 9.dxe5 g6 10.Be2 Bg7 11.h3 f5 12.Nd4 Bxe5 13.Bxh5 0-0 14.0-0 gxh5 15.Qxh5
  +/=  (0.33)   Depth: 28/35   00:01:06  80780kN
1.d4
  +/=  (0.41 ++)   Depth: 29/35   00:01:11  87148kN
1.d4 Nf6
  =  (0.26 --)   Depth: 29/38   00:01:51  136MN
1.d4 Nf6 2.Nf3 e6 3.c4 b6 4.Nc3 Bb4 5.Bg5 h6 6.Bh4 Bb7 7.e3 g5 8.Bg3 Ne4 9.Qc2 f5 10.Bd3 Bxc3+ 11.bxc3 0-0 12.0-0 c5 13.Nd2 Nxg3 14.fxg3 Nc6 15.e4 cxd4 16.exf5 dxc3 17.Qxc3 Qf6 18.Qxf6 Rxf6 19.fxe6
  +/=  (0.35)   Depth: 29/41   00:02:09  157MN

Stockfish takes 20.9% more time to reach the same position than Brainfish.
Parent - By leavenfish (***) [us] Date 2018-02-08 03:21
Peter thanks - I completely forgot their might be a Brainfish compile!
Parent - By Dragon Mist (****) [hr] Date 2018-02-08 19:26
Large Pages will do that, yes. Folks should note that using LP will put that much more stress to their CPUs (the same as additional overclock) i.e. higher temps.
Up Topic The Rybka Lounge / Computer Chess / Stockfish 9

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill