Stockfish 12 has arrived!

[mwyoung]

Is Hyperthreading preferable in NNUE?

Good question, most likely yes, As always it needs to be tested on your system.

The more physical cores on your system. Usually the more you need to watch out for over threading.

8 cores could work well with 16 threads. But if you have 64 cores. It may not be the best going to 128 threads...

[Jamal Bubker]

Thanks Daniel for reporting and thanks to Stockfish Team !!

[quote=daniel71 post_id=858935 time=1599080069 user_id=4085]
Let the testing begin!
https://abrok.eu/stockfish/
[/quote]

[nnnnnnnn]

Stockfish 12 still cannot solve the win study [d]2R2K2/p3p3/3p4/p2P4/P7/4p3/6pp/6nk w - - 0 1 within a couple of minutes (4 cores, Mac, 8GB hash via brew). Even after the introductory moves 1. Rc1 e2, Stockfish 12 did not seem to see the winning move 2 Rb1! after a while.

I wish these engines were tuned better for the analysis of studies.

[lkaufman]

My final result (SF 12 vs 11, 1 thread, 2' + 1" adapted by CCRL rules for hardware) was 180 wins, 19 losses, 301 draws, which is +116 elo. Curiously SF 12 actually got a slightly higher average depth (.23 ply) despite only about 76% NPS.

I ran the same test on four threads, and got +142 -9 = 275 for 112 elo after 426 games, which is a smaller elo drop from 1 thread than would be expected normally. So it's probably fair to say it's about a hundred elo stronger than SF11 under more typical fast rapid time controls, but not 100 elo stronger than the strongest nonNNUE SF (July 31).

[peter]

Stockfish 12 still cannot solve the win study [d]2R2K2/p3p3/3p4/p2P4/P7/4p3/6pp/6nk w - - 0 1 within a couple of minutes (4 cores, Mac, 8GB hash via brew). Even after the introductory moves 1. Rc1 e2, Stockfish 12 did not seem to see the winning move 2 Rb1! after a while.

I wish these engines were tuned better for the analysis of studies.

2R2K2/p3p3/3p4/p2P4/P7/4p3/6pp/6nk w - - 0 1

Analysis by Cfish 010920 64 POPCNT NUMA:

Code: Select all

1.Rc1 a6 2.Kxe7 e2 3.Kxd6 e1N 4.Rxe1 
  =  (0.00)   Depth: 7/7   00:00:00  8kN
...
1.Rc1 
  =  (0.08 ++)   Depth: 60/29   00:00:16  187MN
...
  +/-  (0.71 ++)   Depth: 60/31   00:00:23  299MN
1.Rc1 
  +/-  (1.03 ++)   Depth: 60/31   00:00:28  353MN
...
1.Rc1 
  +-  (4.87 ++)   Depth: 60/71   00:00:35  452MN
1.Rc1 
  +-  (6.37 ++)   Depth: 60/71   00:00:35  453MN
1.Rc1 
  +-  (8.26 ++)   Depth: 60/71   00:00:35  453MN
1.Rc1 
  +-  (10.66 ++)   Depth: 60/71   00:00:35  454MN
1.Rc1 
  +-  (13.67 ++)   Depth: 60/71   00:00:35  454MN
1.Rc1 e5 2.dxe6 e2 3.Rb1 d5 4.e7 d4 5.e8B d3 6.Bh5 a6 7.Ra1 e1Q 8.Rxe1 d2 9.Ra1 d1R 10.Bxd1 Nh3 11.Bc2+ Ng1 12.Bb1 Nf3 13.Be4+ Ng1 14.Ra2 Nf3 15.Bxf3 Kg1 16.Rxg2+ Kf1 17.Rxh2 Kg1 18.Rh5 Kf2 19.Bb7 Ke2 20.Rxa5 Kd3 21.Rxa6 Kd4 22.Ke7 Kc4 23.Rb6 Kc5 24.a5 Kd4 25.Bc6 Kc5 26.Be4 Kd4 27.Kf8 Ke3 28.Rd6 Kf4 29.Ra6 Kg5 30.Ke7 Kg4 31.Rb6 Kg3 32.Kf7 
  +-  (13.77)   Depth: 60/71   00:00:35  454MN
...
1.Rc1 
  +-  (84.33 ++)   Depth: 61/68   00:00:37  490MN
1.Rc1 
  +-  (85.90 ++)   Depth: 61/68   00:00:37  495MN
1.Rc1 
  +-  (90.91 ++)   Depth: 61/68   00:00:37  495MN
1.Rc1 
  +-  (#43 ++)   Depth: 61/68   00:00:37  495MN
1.Rc1 e2 2.Rb1 e6 3.dxe6 d5 4.e7 d4 5.e8B d3 6.Bh5 a6 7.Ra1 e1R 8.Rxe1 d2 9.Ra1 d1R 10.Bxd1 Nh3 11.Bc2+ Ng1 12.Bb1 Nf3 13.Be4+ Ng1 14.Ra2 Nh3 15.Rxg2 Nf4 16.Rg5+ Ng2 17.Rg3 Kg1 18.Rxg2+ Kf1 19.Rxh2 Kg1 20.Rg2+ Kh1 21.Ra2+ Kg1 22.Ke7 Kf1 23.Kd6 Kg1 24.Kc5 Kf1 25.Kd4 Kg1 26.Ke3 Kf1 27.Ra1# 
  +-  (#27)   Depth: 61/68   00:00:37  495MN
...
1.Rc1 e2 2.Rb1 e6 3.dxe6 d5 4.e7 d4 5.e8B d3 6.Bh5 a6 7.Ra1 e1R 8.Rxe1 d2 9.Ra1 d1R 10.Bxd1 Nh3 11.Bc2+ Ng1 12.Bb1 Nf3 13.Be4+ Ng1 14.Ra2 Nh3 15.Rxg2 Nf4 16.Rg6+ Ng2 17.Ke7 Kg1 18.Rxg2+ Kf1 19.Rxh2 Kg1 20.Ra2 Kf1 21.Kd6 Kg1 22.Kc5 Kf1 23.Kd4 Kg1 24.Ke3 Kf1 25.Ra1# 
  +-  (#25)   Depth: 66/54   00:00:40  533MN
...
1.Rc1 e2 2.Rb1 e6 3.dxe6 d5 4.e7 d4 5.e8B d3 6.Bh5 a6 7.Ra1 e1R 8.Rxe1 d2 9.Ra1 d1R 10.Bxd1 Nh3 11.Bc2+ Ng1 12.Bb1 Nf3 13.Be4+ Ng1 14.Ra2 Nh3 15.Rxg2 Nf4 16.Rg4+ Ng2 17.Ke7 Kg1 18.Rxg2+ Kf1 19.Rxh2 Kg1 20.Ra2 Kf1 21.Kd6 Kg1 22.Kc5 Kf1 23.Kd4 Kg1 24.Ke3 Kf1 25.Ra1# 
  +-  (#25)   Depth: 100/50   00:01:08  971MN

24 Threads of 12x3GHz Xeon X5670 CPU (SSE4.1-popcnt- compile) 32 G hash,
Dietrich Kappe's NightNurse 0.2- net.
Use NNUE Pure

Even the right DTM within 40".

regards

[Jhoravi]

How much processing power does SF11 need to add in order to gain 120 ELO at the level of NNUE? Like 2X?

[Leo]

It seems very powerful.

[lkaufman]

How much processing power does SF11 need to add in order to gain 120 ELO at the level of NNUE? Like 2X?

I think it needs somewhere between 3x and 4x the speed to be equal, maybe I'll measure that at some point.

[Modern Times]

How does Stockfish 12 with the NNUE perform at chess960 ? Is the NNUE enabled when the UCI_Chess960 parameter is enabled ?

[OliverBr]

I am having problems to compile the master of https://github.com/official-stockfish/Stockfish both on Linux and MacOSX:

Code: Select all

make stockfish ARCH=x86-64-avx2
...
/tmp/cc3jJ8IN.s: Assembler messages:
/tmp/cc3jJ8IN.s:9: Error: file not found: nn-308d71810dff.nnue
make[1]: *** [/tmp/ccqKQCil.mk:2: /tmp/ccx6iVxC.ltrans0.ltrans.o] Error 1
lto-wrapper: fatal error: make returned 2 exit status
compilation terminated.
/usr/bin/ld: error: lto-wrapper failed
collect2: error: ld returned 1 exit status

I guess something went wrong.. nn? Who can help?