Chess2u
Would you like to react to this message? Create an account in a few clicks or log in to continue.

Chess2uLog in

Stockfish

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
tks..Quo... wink

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
I find the newest engine to be slower in Nodes Per Second than the most recent update PRIOR to changing eval cache instead of TT to store nodes.

I was hoping for much more since it posted a +18 elo on the GitHub web site.

descriptionStockfish - Page 8 EmptyLarge Pages

more_horiz
Why does Stockfish not support "Large Pages"?

On other engines this results in a speed bump of > +10%.

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
FlyingDutchman wrote:
Why does Stockfish not support "Large Pages"?

On other engines this results in a speed bump of > +10%.

+1

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
[pgn][Event "Blitz 5 minutes"]
[Site "Engine Match"]
[Date "2012.12.07"]
[Round "?"]
[White "Sfish 121204Qi x64"]
[Black "Stockfish 2.3.1 JA 64bit"]
[Result "1-0"]
[ECO "C11"]
[Annotator "0.12;0.04"]
[PlyCount "139"]
[TimeControl "300"]

1. e4 {B 0} e6 {B 0
} 2. d4 {B 0} d5 {B 0} 3. Nc3 {B 0} Nf6 {B 0} 4. e5 {B 0} Nfd7 {B 0} 5. f4 {B 0
} c5 {B 0} 6. Nf3 {B 0} Nc6 {B 0} 7. Be3 {B 0} cxd4 {B 0} 8. Nxd4 {B 0} Bc5 {
B 0} 9. Qd2 {B 0} O-O {B 0} 10. O-O-O {B 0} a6 {B 0 Both last book move} 11.
Nb3 {0.12/21 6} Be7 {0.04/19 5 (Bxe3)} 12. Kb1 {0.00/19 7 (h4)} b5 {0.00/19 6}
13. h4 {0.00/19 8 (g4)} Nb6 {-0.08/22 23 (Qc7)} 14. Qf2 {-0.08/20 13} Nc4 {0.
00/21 5 (Rb8)} 15. Bc5 {0.08/21 5} Bxc5 {0.04/21 5} 16. Nxc5 {0.04/21 8} Rb8 {
0.00/22 5} 17. g3 {0.16/21 6} Qa5 {0.08/22 21 (h6)} 18. h5 {0.00/21 20} h6 {0.
00/19 5} 19. Be2 {0.00/22 6 (g4)} Bb7 {-0.16/20 7 (Na3+)} 20. Rhf1 {0.00/19 15
(Bxc4)} Rfe8 {0.00/19 7 (Ba8)} 21. Bd3 {0.00/18 6} Ba8 {0.00/20 6} 22. Nb3 {0.
00/22 7} Qc7 {0.20/20 12} 23. g4 {0.08/22 6} Rec8 {0.24/19 4 (Nb4)} 24. Ne2 {
0.20/17 6 (Rg1)} a5 {0.00/18 10} 25. Nbd4 {0.60/18 6} Qe7 {0.20/20 8} 26. Nxc6
{0.80/20 10 (g5)} Bxc6 {0.72/20 8} 27. Nd4 {0.48/20 12} Qb4 {0.92/20 9} 28. c3
{0.56/22 4} Qe7 {0.64/20 12} 29. g5 {0.21/21 15} hxg5 {0.64/20 4} 30. fxg5 {0.
80/19 4} Be8 {0.60/20 4} 31. h6 {0.52/20 7} g6 {0.52/21 4} 32. Ka1 {0.44/20 13
(Bxc4)} b4 {0.08/19 5} 33. cxb4 {0.44/19 4 (Bxc4)} Rxb4 {0.40/21 4} 34. Bxc4 {
0.36/20 4 (b3)} Rcxc4 {0.36/20 6} 35. Rg1 {0.24/20 3} a4 {0.04/20 5 (Rb8)} 36.
a3 {0.28/19 3 (Kb1)} Rb8 {0.32/20 4 (Rb7)} 37. Qd2 {0.16/19 2 (Qe3)} Qb7 {0.32/
23 4 (Qc7)} 38. Rh1 {0.24/21 2} Qe7 {0.32/24 4 (Qc7)} 39. Rc1 {0.36/23 3} Rbc8
{0.40/26 12} 40. Rxc4 {0.36/26 2} Rxc4 {0.68/24 3} 41. Kb1 {0.44/26 4 (Rc1)}
Rc7 {0.40/28 3 (Kh7)} 42. Qf4 {0.40/26 2} Kh7 {0.40/29 3} 43. Rc1 {0.40/28 4
(Ka1)} Rb7 {0.40/27 3} 44. Rc2 {0.40/28 3 (Rc3)} Bd7 {0.40/27 3} 45. Rf2 {0.40/
28 4} Be8 {0.68/28 6} 46. Rc2 {0.40/30 3 (Rf3)} Bd7 {0.00/34 2} 47. Rf2 {0.40/
29 3} Be8 {0.00/34 2} 48. Rg2 {0.40/28 3 (Rf3)} Qd8 {0.68/22 3 (Bd7)} 49. Rc2 {
1.05/23 5 (Rg3)} Qe7 {0.68/24 2} 50. Rc3 {1.45/23 2} Bd7 {1.09/25 6 (Bb5)} 51.
Nf3 {1.73/24 2 (Ka1)} Bb5 {0.12/24 2} 52. Qf6 {1.73/26 2 (Re3)} Qxf6 {1.69/20 2
} 53. exf6 {1.85/12 0} Bc4 {1.93/23 2} 54. Ne5 {2.06/25 1} Rc7 {2.06/26 1} 55.
Kc1 {2.26/27 3 (Kc2)} Kg8 {2.22/25 3} 56. Rc2 {2.26/25 2} Ra7 {2.46/25 2} 57.
Kd2 {3.87/25 5 (Nxc4)} Rb7 {2.42/21 1} 58. Ke3 {3.71/25 1} Rb3+ {3.07/23 4
(Rc7)} 59. Kd4 {3.83/23 1} Rh3 {3.53/24 4 (Rb7)} 60. Kc5 {4.68/20 1 (Nxc4)} Bb3
{4.12/19 1} 61. Rc3 {5.21/25 1 (Rc1)} Rh2 {5.29/23 1} 62. Kd6 {5.89/25 1} Bc4 {
5.33/24 1} 63. Nxc4 {6.14/25 1} dxc4 {5.45/24 1} 64. Rxc4 {6.14/12 0} Rxb2 {5.
37/24 1} 65. Rc8+ {6.14/13 0} Kh7 {5.81/24 1} 66. Rc7 {6.38/24 1} Kg8 {6.26/24
1} 67. h7+ {6.54/25 1} Kxh7 {7.55/24 1} 68. Rxf7+ {6.54/12 0} Kg8 {8.24/25 1}
69. Rg7+ {6.54/23 1} Kf8 {10.91/25 2} 70. Rxg6 {7.43/21 1} 1-0[/pgn]

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Stockfish source - Dec 08, 2012

- Micro-optimize color_of()

https://github.com/mcostalba/Stockfish/commits/master

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Where can I contact Mr. Tom Romstad regarding which version of Stockfish he wants me to use for the Supertournament?

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Castle00 wrote:
Where can I contact Mr. Tom Romstad regarding which version of Stockfish he wants me to use for the Supertournament?

Here:
http://support.stockfishchess.org/discussion/new

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Stockfish source - Dec 10, 2012

- Merge branch 'eval_cache'

Merge branch 'eval_cache'

Unusually good result. Defenitly needs further verifications.

After 2160 games at 15"+0.05
Mod vs Orig 486 - 367 - 1307 ELO +19

bench: 6261882

Stockfish source - Dec 09, 2012

- Fix comparison with alpha, not beta
- Ensure valueLower <= valueUpper
- Store distinct upper and lower bound scores

https://github.com/mcostalba/Stockfish/commits/master

Good ! cool

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Big Test!

Time Control: 3 Minutes
1500 Games

Stockfish 2.3.1 JA 64-bit vs New-Stockfish-Version (any suggestion?)

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Vladimir wrote:
Big Test!

Time Control: 3 Minutes
1500 Games

Stockfish 2.3.1 JA 64-bit vs New-Stockfish-Version (any suggestion?)


A great idea!

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Quocvuong82 wrote:
Vladimir wrote:
Big Test!

Time Control: 3 Minutes
1500 Games

Stockfish 2.3.1 JA 64-bit vs New-Stockfish-Version (any suggestion?)


A great idea!

any suggestion? new build!?

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Stockfish source - Dec 15, 2012

- Don't need to check for bestValue < beta to split
- Remove an extra semicolon
- Revert store of distinct upper and lower bounds

Revert store of distinct upper and lower bounds

Grand totals after 10040 games (crashes: 0) for tt_both

master_9edc7 - 6a93488_6a934: 1756 - 1688 - 6596 ELO +2 (+- 2.7)

Confirmed by test of Gary:

After 8680 games:
ELO: 0.80 +- 99%: 9.62 95%: 7.31
LOS: 65.38%
Wins: 1288 Losses: 1268 Draws: 6130

https://github.com/mcostalba/Stockfish/commits/master

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Latest source compiled as of today with Intel Compiler XE 13 for Intel chips only. Squeezed the last 5kn/s performance out with an Intel PGO build that I let run and optimize for 2 hours.

Intel x64 SSE4.2 w/ POPCNT + PGO
DOWNLOAD: mediafire.com ?439sqblqmvmdnfg

Searches 5,000 nodes faster than non-PGO on my machine Intel 2600k 4.0 Ghz

Non-PGO = 1661911 nodes/sec avg
PGO = 1693879 nodes/sec avg

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Thanks deebix, what about Stockfish 64bit (not-SSE4.2) ?

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
I will compile in a little bit for non sse4.2. I believe technically the compile should still work on your machine even if you do not have an intel CPU.

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
any chance to leave a copilation amd?

descriptionStockfish - Page 8 EmptyStockfish

more_horiz
could you please compile a 32 bit version!...(deebix)

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Stockfish source - Dec 25, 2012

- Introduce Null Threat extension:
In case of null search at low depths returns a fail low
due to a threat then, rather than return beta-1 (to cause
a re-search at full depth in the parent node), we set a flag
threatExtension = true (false by default) that will cause
moves that prevent the threat to be extended of one ply
in the following search.

Idea and patch is by Lucas Braesch.

Lucas also did the tests:
1500 games in 5"+0.05":
SF_threatExtension vs SF_20121222: 366 - 331 - 803 [51.2%] LOS=90.8%

3000 games in 10"+0.1":
SF_threatExtension vs SF_20121222: 610 - 559 - 1831 [50.8%] LOS=93.2%

Tests confirmed by Gary after 10570 games,
ELO: 2.79 +- 99%: 8.72 95%: 6.63
LOS: 94.08%
Wins: 1523 Losses: 1438 Draws: 7607

And finally by me at 15"+0.05, single thread, 3824 games
threatExtension vs master 768 - 692 - 2364 +7 ELO

bench 4918443

- Fix a couple of extra spaces
- Retire Position::in_check()
- Retire Position::move_is_legal()
- Small tweak in is_pseudo_legal()

https://github.com/mcostalba/Stockfish/commits/master

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Stockfish (24-12-2012)

Linux x64
Linux x64 for modern computers
Linux x64 for modern computers sse4.2
Windows x64
Windows x64 for modern computers
Windows x64 for modern computers sse4.2

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
so many version of stockfish pls stop publishing too many version only tested strong and can match houdini 3 . too many all weak version non sense what a waste

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Stockfish (26-12-2012)

Linux x64
Linux x64 for modern computers
Linux x64 for modern computers sse4.2
Windows x64
Windows x64 for modern computers
Windows x64 for modern computers sse4.2

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
there is no version for 32 bits ?

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Stockfish Development Versions are build automatically if there are changes on the master branch in the git repository (https://github.com/mcostalba/Stockfish).
They are compiled with gcc 4.7/mingw on Ubuntu 12.10.

Windows x64
Windows x64 SSE4.2
Linux x64
Linux x64 SSE4.2

http://abrok.eu/stockfish/

descriptionStockfish - Page 8 Emptynewest version appears to be a regression

more_horiz

Code:

Games Completed = 1027 of 50000 (Avg game length = 13.611 sec)
Settings = RR/32MB/500ms+100ms/M 1000cp for 12 moves, D 150 moves/
Time = 7441 sec elapsed, 354818 sec remaining
 1.  Ble                         537.5/1027   236-188-603     (L: m=24 t=0 i=0 a=164)   (D: r=483 i=85 f=15 s=0 a=20)   (tpm=97.6 d=15.21 nps=3807996)
 2.  latest commit               489.5/1027   188-236-603     (L: m=36 t=0 i=0 a=200)   (D: r=483 i=85 f=15 s=0 a=20)   (tpm=97.7 d=15.00 nps=3306742)

after 1027 games, the latest version of Stockfish still loses badly to a much earlier version. I believe that tampering with Transp. Table appears to have made the newer version much slower. Check out the NPS 3807996 vs. 3306742. The previous version with a lazy eval setting is 500,000 nodes faster with no loss in tactical strength

+9 elo lol

descriptionStockfish - Page 8 EmptyRe: Stockfish

more_horiz
Permissions in this forum:
You cannot reply to topics in this forum