New 64-bit Stockfish 1.5.1 (oct. 10) behaves same as 32-bit

[ernest]

Indeed, the new stockfish_151_intel_x64.exe from october 12 is about 4% faster than the previous stockfish_151_x64_ja.exe from october 10
(according to the kn/s in infinite analysis from the start_position, 1 thread),
but no longer behaves the same as the 32-bit version.

Maybe this is normal, and Jim could compile a new 32-bit version, which would also be faster than the previous 32-bit one, and would behave the same way as the new 64-bit.

[Jim Ablett]

Indeed, the new stockfish_151_intel_x64.exe from october 12 is about 4% faster than the previous stockfish_151_x64_ja.exe from october 10
(according to the kn/s in infinite analysis from the start_position, 1 thread),
but no longer behaves the same as the 32-bit version.

Maybe this is normal, and Jim could compile a new 32-bit version, which would also be faster than the previous 32-bit one, and would behave the same way as the new 64-bit.

I haven't been able to compile a faster 32 bit version than original which used Msvc 2005,
although I haven't tried using latest Msvc 2010 on the command line yet, which Mincho says
can use pgo (profile guided optimization). I will try this later today.
This wont behave the same as Intel 64 bit though, as mentioned earlier Intel & Msvc
implement move sorting functions which behave slightly differently.

Jim.

[Mincho Georgiev]

Personally, i don't see any improvement in vs10's compiler. I'm really sorry to say it,but so far at least for me, the icc gives better results. I hope you prove me wrong, really.

[Uri Blass]

I've got a pretty decent speedup with the new version Intel compiler I updated today.
You can download this new compile here if you wish >

http://www.mediafire.com/?zwmeijm33yx

Jim.

Hi Jim,

this is a 3.5% speed up !!

This is really a nice improvment...for sure it translates in some elo points more.

Thanks a lot
Marco


BTW have you tried with MSVC 2010 ? On my PC it is the fastest binary I can get....but slower then yours of course

I did not try MSVC 2010 but it is only 2009 so I wonder why the name 2010 for something that you can use in 2009

It seems to me a bad name.
Uri

[ernest]

Hi Jim,
Actually, the new stockfish_151_intel_x64.exe from october 12 behaves the same as the stockfish_151_x64_ja.exe from october 8 (and is 4% faster).

Code: Select all

Analysis by Stockfish 1.5.1 64bit:   (12/10/2009)
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.0-0 Nf6 5.Nc3 0-0 6.d3 d6 7.Na4 Na5 8.Nxe5 Nxc4 9.Nxc4 Bg4 10.Qe1 
  +/=  (0.28)   Depth: 18   00:00:32  44738kN
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.0-0 Nf6 5.Nc3 0-0 6.d3 d6 7.Bg5 Be6 8.Nd5 Bxd5 9.Bxd5 h6 10.Bxc6 bxc6 
  +/=  (0.28)   Depth: 19   00:01:03  87897kN
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.d3 Bg4 6.Nbd2 Qe7 7.Nc4 0-0-0 8.0-0 f6 9.Be3 Kb8 10.h3 Be6 
  +/=  (0.28)   Depth: 20   00:02:05  175mN
Analysis by Stockfish 1.5.1 JA 64bit:   (08/10/2009)
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.0-0 Nf6 5.Nc3 0-0 6.d3 d6 7.Na4 Na5 8.Nxe5 Nxc4 9.Nxc4 Bg4 10.Qe1 
  +/=  (0.28)   Depth: 18   00:00:33  44738kN
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.0-0 Nf6 5.Nc3 0-0 6.d3 d6 7.Bg5 Be6 8.Nd5 Bxd5 9.Bxd5 h6 10.Bxc6 bxc6 
  +/=  (0.28)   Depth: 19   00:01:05  87897kN
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.d3 Bg4 6.Nbd2 Qe7 7.Nc4 0-0-0 8.0-0 f6 9.Be3 Kb8 10.h3 Be6 
  +/=  (0.28)   Depth: 20   00:02:10  175mN

Only the 64-bit stockfish_151_x64_ja.exe from october 10 behaves the same as the 32-bit.

Code: Select all

Analysis by Stockfish 1.5.1 JA 64bit:  (10/10/2009)
1.e4 e6 2.Nf3 d5 3.e5 c5 4.Nc3 Bd7 5.Bb5 a6 6.Bxd7+ Kxd7 7.0-0 Nc6 8.d3 Be7 9.Qe2 Nh6 10.Bxh6 gxh6 
  +/=  (0.40)   Depth: 18   00:00:39  52836kN
1.e4 e6 2.Nf3 d5 3.e5 Nh6 4.d4 c5 5.Bd3 cxd4 6.0-0 Nc6 7.Bb5 Nf5 8.g4 Nh6 9.Nxd4 Bd7 10.Nxc6 Bxc6 11.Bxc6+ bxc6 
  +/=  (0.36)   Depth: 19   00:01:17  104mN
1.e4 e6 2.Nf3 d5 3.e5 Nh6 4.d4 c5 5.Bd3 cxd4 6.0-0 Nc6 7.Bb5 Nf5 8.g4 Nh4 9.Nxd4 Bd7 10.Nxc6 bxc6 11.Bd3 
  +/=  (0.32)   Depth: 20   00:02:31  202mN
Analysis by Stockfish 1.5.1 JA:   32-bit
1.e4 e6 2.Nf3 d5 3.e5 c5 4.Nc3 Bd7 5.Bb5 a6 6.Bxd7+ Kxd7 7.0-0 Nc6 8.d3 Be7 9.Qe2 Nh6 10.Bxh6 gxh6 
  +/=  (0.40)   Depth: 18   00:00:48  52836kN
1.e4 e6 2.Nf3 d5 3.e5 Nh6 4.d4 c5 5.Bd3 cxd4 6.0-0 Nc6 7.Bb5 Nf5 8.g4 Nh6 9.Nxd4 Bd7 10.Nxc6 Bxc6 11.Bxc6+ bxc6 
  +/=  (0.36)   Depth: 19   00:01:35  104mN
1.e4 e6 2.Nf3 d5 3.e5 Nh6 4.d4 c5 5.Bd3 cxd4 6.0-0 Nc6 7.Bb5 Nf5 8.g4 Nh4 9.Nxd4 Bd7 10.Nxc6 bxc6 11.Bd3 
  +/=  (0.32)   Depth: 20   00:03:06  202mN

[Spock]

Crikey, I'm losing track of all these x64 compiles
For testing purposes, does it matter what I use ?

[Spock]

Hmmmm SF blocks on the last move before time control and it blocks when he has still many seconds on the clock.

Sf also rushes the last moves like the avaiable time is much less then what actually is. Could you please also post some game won by SF, just to see if it properly used all the available time.

Sorry, what is the GUI used ?


BTW does this laso at fixed time controls as example 4+0 ?

Thanks

Well, it seems Stockfish is just not too happy running under ChessGUI 0.149

The first batch 10 Oct exe - 3 time losses that I already posted
The second batch 12 Oct exe - 4 time losses

Shredder GUI, 10 Oct exe - no time losses and a substantially stronger result

All 3 matches from the same pgn of 50 opening positions, reversed sides.

Perhaps time to upgrade ChessGUI to the latest version, but to avoid time losses looks like I'll use the Shredder GUI for Stockfish for all the UCI vs UCI matches.

[Alexander Schmidt]

Roughly how many ELO stronger would you expect this to be ?

Much stronger.

Code: Select all

    Motor              Punkte       St    Sh    To    Fr    Sm    Gl    Sp    Na    Br    Mo    Fr    Ph    Th    Ha    Jo    Uf    S-B
01: StockFish 1.5.1    104,5/120 ····· 6,0/8 4,5/8 8,0/8 6,0/8 6,5/8 5,0/8 7,0/8 6,5/8 8,0/8 8,0/8 7,5/8 8,0/8 8,0/8 7,5/8 8,0/8  5754,5
02: Shredder Classic 4 81,0/120  2,0/8 ····· 3,5/8 4,0/8 4,0/8 5,5/8 5,0/8 5,5/8 3,5/8 7,5/8 6,0/8 5,5/8 5,0/8 8,0/8 8,0/8 8,0/8  4283,5
03: Toga CMLX 1.4.5e   80,0/120  3,5/8 4,5/8 ····· 4,0/8 6,5/8 4,5/8 5,5/8 5,5/8 3,0/8 5,5/8 6,0/8 7,0/8 4,5/8 8,0/8 8,0/8 4,0/8  4457,5
04: Fruit 2.3.1        76,0/120  0,0/8 4,0/8 4,0/8 ····· 3,5/8 4,5/8 5,0/8 6,0/8 5,5/8 5,5/8 5,0/8 7,0/8 7,0/8 7,0/8 6,0/8 6,0/8  4005,2
05: Smaug 2.2.1        74,5/120  2,0/8 4,0/8 1,5/8 4,5/8 ····· 3,5/8 5,5/8 3,5/8 4,5/8 6,0/8 5,0/8 7,0/8 6,0/8 7,5/8 6,0/8 8,0/8  3902,0
06: Glaurung 2.2       74,0/120  1,5/8 2,5/8 3,5/8 3,5/8 4,5/8 ····· 5,5/8 5,0/8 5,5/8 6,5/8 7,0/8 6,0/8 5,5/8 7,0/8 5,0/8 5,5/8  3976,0
07: Spike 1.2          64,0/120  3,0/8 3,0/8 2,5/8 3,0/8 2,5/8 2,5/8 ····· 4,0/8 5,0/8 5,5/8 4,5/8 3,5/8 6,5/8 5,0/8 7,0/8 6,5/8  3426,7
08: Naum 2.0           63,5/120  1,0/8 2,5/8 2,5/8 2,0/8 4,5/8 3,0/8 4,0/8 ····· 4,5/8 3,5/8 7,0/8 5,5/8 5,5/8 5,0/8 7,0/8 6,0/8  3315,5
09: Bright 0.4a        62,5/120  1,5/8 4,5/8 5,0/8 2,5/8 3,5/8 2,5/8 3,0/8 3,5/8 ····· 5,0/8 4,0/8 6,5/8 4,5/8 7,0/8 5,0/8 4,5/8  3425,0
10: Movei 0.08.438     49,5/120  0,0/8 0,5/8 2,5/8 2,5/8 2,0/8 1,5/8 2,5/8 4,5/8 3,0/8 ····· 4,0/8 4,0/8 4,5/8 5,5/8 6,0/8 6,5/8  2464,2
11: Frenzee 02.2008    49,0/120  0,0/8 2,0/8 2,0/8 3,0/8 3,0/8 1,0/8 3,5/8 1,0/8 4,0/8 4,0/8 ····· 5,0/8 6,0/8 2,5/8 5,5/8 6,5/8  2502,0
12: Pharaon 3.51       43,5/120  0,5/8 2,5/8 1,0/8 1,0/8 1,0/8 2,0/8 4,5/8 2,5/8 1,5/8 4,0/8 3,0/8 ····· 5,0/8 5,0/8 5,0/8 5,0/8  2208,7
13: The Baron 2.2.3    37,0/120  0,0/8 3,0/8 3,5/8 1,0/8 2,0/8 2,5/8 1,5/8 2,5/8 3,5/8 3,5/8 2,0/8 3,0/8 ····· 4,0/8 2,5/8 2,5/8  2114,7
14: Hamsters 0.7.1     36,0/120  0,0/8 0,0/8 0,0/8 1,0/8 0,5/8 1,0/8 3,0/8 3,0/8 1,0/8 2,5/8 5,5/8 3,0/8 4,0/8 ····· 4,5/8 7,0/8  1676,5
15: Jonny 2.83         33,0/120  0,5/8 0,0/8 0,0/8 2,0/8 2,0/8 3,0/8 1,0/8 1,0/8 3,0/8 2,0/8 2,5/8 3,0/8 5,5/8 3,5/8 ····· 4,0/8  1699,7
16: Ufim 8.02          32,0/120  0,0/8 0,0/8 4,0/8 2,0/8 0,0/8 2,5/8 1,5/8 2,0/8 3,5/8 1,5/8 1,5/8 3,0/8 5,5/8 1,0/8 4,0/8 ·····  1748,5

Code: Select all

   Motor            Punkte       Ry      Sh      Na      St      To      Gl      Lo      Fr    S-B
1: Rybka 3          80,5/112 ·······  8,0/16 10,5/16 10,5/16 14,5/16 11,5/16 12,5/16 13,0/16  4058,0
2: Shredder 12      72,5/112  8,0/16 ·······  6,5/16 11,0/16 12,5/16  8,0/16 12,5/16 14,0/16  3683,5
3: Naum 4           70,5/112  5,5/16  9,5/16 ·······  5,0/16 13,5/16 12,5/16 11,5/16 13,0/16  3451,7
4: StockFish 1.5.1  68,0/112  5,5/16  5,0/16 11,0/16 ·······  8,5/16 12,0/16 12,5/16 13,5/16  3374,5
5: Toga CMLX 1.4.5e 44,5/112  1,5/16  3,5/16  2,5/16  7,5/16 ······· 13,0/16  9,0/16  7,5/16  2172,2
6: Glaurung 2.2     39,0/112  4,5/16  8,0/16  3,5/16  4,0/16  3,0/16 ·······  8,5/16  7,5/16  2180,0
7: Loop 10.32f      38,0/112  3,5/16  3,5/16  4,5/16  3,5/16  7,0/16  7,5/16 ·······  8,5/16  1992,2
8: Fruit 2.3.1      35,0/112  3,0/16  2,0/16  3,0/16  2,5/16  8,5/16  8,5/16  7,5/16 ·······  1762,7

4min/game+2sec/move

[mcostalba]

Well, it seems Stockfish is just not too happy running under ChessGUI 0.149

I use Fritz 11 GUI for testing and never had problems, I don't have ChessGUI....I could download 0.149 and do some debug, but this week I am quite busy, so probably it is _another_ job queued up in the TODO list

[Spock]

Well, it seems Stockfish is just not too happy running under ChessGUI 0.149

I use Fritz 11 GUI for testing and never had problems, I don't have ChessGUI....I could download 0.149 and do some debug, but this week I am quite busy, so probably it is _another_ job queued up in the TODO list

Current version is 0.167, you won't be able to download the earlier one.
I could send you the entire 100 games if you give me an email address, and you could check a handful of the won games and see if Glaurung was also having issues, as you requested. In any event, I'll run the UCI vs UCI games in Shredder GUI now, and update to the latest ChessGUI to run the UCI vs Winboard ones.