Designing an analysis friendly Stockfish?

Code, algorithms, languages, construction...
Jeremy Bernstein
Site Admin
Posts: 1226
Joined: Wed Jun 09, 2010 7:49 am
Real Name: Jeremy Bernstein
Location: Berlin, Germany
Contact:

Re: Designing an analysis friendly Stockfish?

Post by Jeremy Bernstein » Thu Feb 24, 2011 5:01 pm

keoki010 wrote:
Jeremy Bernstein wrote:
Bildschirmfoto 2011-02-24 um 07.55.04.png
2d easily beat 2c in 3/3 however it still had 1 time loss. it was +5 =4-1 for a 10 game match 3/3
10 games is pretty much random, I think -- the only change between 2d and 2c is a correction of the amount of time spent generating pvs from tablebase hits at the root. That means, only when 5 or fewer pieces are on the board would there be any change, and the only change would be that it returns a shorter pv in less time when there isn't much time left.

I'm surprised that there was a time loss from 2d, though -- there's no reason I can think of that would cause it (if that sort of time loss doesn't happen sometimes to stockfish-ja). You can send me the pgn and/or another other output if you want me to take a look.

Thanks
Jeremy

keoki010
Posts: 32
Joined: Mon Aug 16, 2010 4:53 pm

Re: Designing an analysis friendly Stockfish?

Post by keoki010 » Thu Feb 24, 2011 7:07 pm

Jeremy Bernstein wrote:
keoki010 wrote:
Jeremy Bernstein wrote:
Bildschirmfoto 2011-02-24 um 07.55.04.png
2d easily beat 2c in 3/3 however it still had 1 time loss. it was +5 =4-1 for a 10 game match 3/3
10 games is pretty much random, I think -- the only change between 2d and 2c is a correction of the amount of time spent generating pvs from tablebase hits at the root. That means, only when 5 or fewer pieces are on the board would there be any change, and the only change would be that it returns a shorter pv in less time when there isn't much time left.

I'm surprised that there was a time loss from 2d, though -- there's no reason I can think of that would cause it (if that sort of time loss doesn't happen sometimes to stockfish-ja). You can send me the pgn and/or another other output if you want me to take a look.

Thanks
Jeremy
I think the problem may be Arena. Looking at some of the games it seems that it sometimes will call a time loss with double digit seconds left on the clock.

Jeremy Bernstein
Site Admin
Posts: 1226
Joined: Wed Jun 09, 2010 7:49 am
Real Name: Jeremy Bernstein
Location: Berlin, Germany
Contact:

Re: Designing an analysis friendly Stockfish?

Post by Jeremy Bernstein » Fri Feb 25, 2011 10:04 am

Final stats. Once again, basically a draw.
Engine                          Score                                                                                                                                                                                                          St                                                                                                                                                                                                       St                                                                                                                                                                                                       St    S-B
Stockfish-201-64-ja             206,5/400 ········································································································································································································ ===01=1===1==1000==01==1=====0=10=111==0==1==1=0==1===1==0=1==00=11=00====1=====0==010=0=1=======0======1=1====1=111111====1===1===1===00=====0=01==0=0==1==11====0==1========0=11=0====0==1=1==0======1 =1=001=====11===1===1=0==0=00==0=====1========0==0=011=1===11==010=====0==========0==0====0=====1=====1============1=1==110====0111=0====0=========0====0==11=0======00=1=1==01==0==1=1=1==1=0=0==0==01=  40660,
Stockfish_201_PA_GTB_Gran2d_x64 202,5/400 ===10=0===0==0111==10==0=====1=01=000==1==0==0=1==0===0==1=0==11=00=11====0=====1==101=1=0=======1======0=0====0=000000====0===0===0===11=====1=10==1=1==0==00====1==0========1=00=1====1==0=0==1======0 ········································································································································································································ 0=11==1=========1011=1===1=0====0==0=======10=0111==1=0=01=====10=======10=======1=0=0=1==11=0=101011=1===1=0=11==01===1=====1=1=101=1===1===101=1=0====0=1==0==01==00=========0111======11===1====00100  40134,
Stockfish_201_PA_GTB_Gran2c_x64 191,0/400 =0=110=====00===0===0=1==1=11==1=====0========1==1=100=0===00==101=====1==========1==1====1=====0=====0============0=0==001====1000=1====1=========1====1==00=1======11=0=0==10==1==0=0=0==0=1=1==1==10= 1=00==0=========0100=0===0=1====1==1=======01=1000==0=1=10=====01=======01=======0=1=1=0==00=1=010100=0===0=1=00==10===0=====0=0=010=0===0===010=0=1====1=0==1==10==11=========1000======00===0====11011 ········································································································································································································  39075,

600 games played / Tournament is finished
Level: Blitz 0:30/0,5
Hardware: Intel(R) Xeon(R) CPU  W3520 @ 2.67GHz 2659 MHz with 1,0 GB Memory
Operating system: Windows 7 Professional Professional (Build 7600) 64 bit

Damir Desevac
Posts: 57
Joined: Thu Jun 10, 2010 2:58 pm
Real Name: Damir Desevac

Re: Designing an analysis friendly Stockfish?

Post by Damir Desevac » Fri Feb 25, 2011 8:08 pm

Considering the time losses, Stockfish Gran is worse than Rybka... It has basically crap time management.

keoki010
Posts: 32
Joined: Mon Aug 16, 2010 4:53 pm

Re: Designing an analysis friendly Stockfish?

Post by keoki010 » Fri Feb 25, 2011 8:25 pm

eremy here are the results from my last 3/3 Grand slightly better. No time losses I ran it in Auarium 4.0.6 Damn another key gone this keyboard is crap off to the store.
2/25/2011 1:21:14 PM :

Program Elo + - Games Score Av.Op. Draws

1 SF_Gran2d : 2409 46 46 100 52.5 % 2391 55.0 %
2 SF_Gran2c : 2391 46 46 100 47.5 % 2409 55.0 %

keoki010
Posts: 32
Joined: Mon Aug 16, 2010 4:53 pm

Re: Designing an analysis friendly Stockfish?

Post by keoki010 » Fri Feb 25, 2011 8:46 pm

Sorry I looked at the pgn on a diff gui and there are time losses by both 2c and 2d.

Peter C
Posts: 154
Joined: Thu Jun 10, 2010 3:12 am
Real Name: Peter C

Re: Designing an analysis friendly Stockfish?

Post by Peter C » Fri Feb 25, 2011 11:13 pm

If you are playing hyperbullet, you need to adjust the time management settings so it doesn't lose on time. Someone suggested the following settings earlier in this thread and I found them to work quite well:

Minimum Thinking Time = 2
Emergency Base Time = 20
Emergency Move Time = 7

Peter

keoki010
Posts: 32
Joined: Mon Aug 16, 2010 4:53 pm

Re: Designing an analysis friendly Stockfish?

Post by keoki010 » Fri Feb 25, 2011 11:35 pm

Peter C wrote:If you are playing hyperbullet, you need to adjust the time management settings so it doesn't lose on time. Someone suggested the following settings earlier in this thread and I found them to work quite well:

Minimum Thinking Time = 2
Emergency Base Time = 20
Emergency Move Time = 7

Peter
Thanks Peter I'll try them tonight. I don't usually play engine v engine I use them for cc analysis. Trying to help Jeremy and Uly develop a better analyses engine.

Jeremy Bernstein
Site Admin
Posts: 1226
Joined: Wed Jun 09, 2010 7:49 am
Real Name: Jeremy Bernstein
Location: Berlin, Germany
Contact:

Re: Designing an analysis friendly Stockfish?

Post by Jeremy Bernstein » Fri Feb 25, 2011 11:59 pm

Damir Desevac wrote:Considering the time losses, Stockfish Gran is worse than Rybka... It has basically crap time management.
Do you ever formulate sentences that don't piss people off?

None of the changes in these builds affect Stockfish's time management, except that TB access requires some amount of time. However, that time is tracked and used in the same time management calculations as the rest of Stockfish. Beyond that, very few of these games ever get to a point where TB access is an issue at all.

But, as they say, You Have The Source Code. Step up and fix it if you can.

Jeremy

Damir Desevac
Posts: 57
Joined: Thu Jun 10, 2010 2:58 pm
Real Name: Damir Desevac

Re: Designing an analysis friendly Stockfish?

Post by Damir Desevac » Sat Feb 26, 2011 10:08 am

Hi Jeremy

Sorry for my choice of words, but apparently this has been said before ( the time losses), and nothing has been done to fix this.
In Fritz Gui the engine losses quite often on time.

Post Reply