Testing Stockfish 11-03-13. 480 Games.

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Rebel, chrisw

Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 9:09 pm

Re: Testing Stockfish 15-03-13. 480 Games.

Post by Tomcass »

A new best result for Stockfish in my test. Now 16-03-13 version. 480 games.

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 4 min +0 sec/ game
No tablebases

Stockfish 160313 – Criter 1.6 +8/=26/-6 21.0 – 19.0 52.50%
Stockfish 160313 – Deep Rybka 4.1 +17/=19/-4 26.5 – 13.5 66.25%
Stockfish 160313 – Houdini 3.0Pro +5/=23/-12 16.5 – 23.5 41.25%

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 min +2 sec/ game
No tablebases

Stockfish 160313 – Criter 1.6 +11/=21/-8 21.5 – 18.5 53.75%
Stockfish 160313 – Deep Rybka 4.1 +11/=24/-5 23.0 – 17.0 57.50%
Stockfish 160313 – Houdini 3.0Pro +1/=30/-9 16.0 – 24.0 40.00%

Overall average with 6 cores : ( 124.5 – 115.5) : 51.87%

240 games: http://www.mediafire.com/?797e377y73iomz6

i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12
Time control: 4 min + 0 sec/ game
Ponder: Off
No tablebases

Stockfish 160313 - Critter 1.6a +6/=25/-9 18.5 – 21.5 46.25%
Stockfish 160313 - Deep Rybka +11/=28/-1 25.0 – 15.0 62.50%
Stockfish 160313 - Houdini3.0Pro +7/=22/-12 18.0 – 22.0 45.00%

i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12
Time control: 2 min + 2 sec/ game
Ponder: Off
No tablebases

Stockfish 160313 - Critter 1.6a +9/=25/-5 21.5 – 18.5 53.75%
Stockfish 160313 - Deep Rybka +7/=27/-6 20.5 – 19.5 51.25%
Stockfish 160313 - Houdini3.0 Pro +4/=21/-15 15.5 – 24.5 38.75%

240 games x 4cores : http://www.mediafire.com/?70z95039crq17i3

Overall average with 4 cores ( 119.0 -121.0) : 49.58%

… and after 480 games: 243.5 – 236.5 : 50.73%

Again a new record for Stockfish, breaking for the first time ever the 50% result in my test.

51.87% against Critter 1.6, 59.69% against Deep Rybka 4 and 40.62% against Houdini 3.0 Pro. (!!)

Well done!.

Regards,

Tom.
gladius
Posts: 568
Joined: Tue Dec 12, 2006 10:10 am
Full name: Gary Linscott

Re: Testing Stockfish 15-03-13. 480 Games.

Post by gladius »

Tomcass wrote:Again a new record for Stockfish, breaking for the first time ever the 50% result in my test.
Great to see! Thanks Tom :). Of course SF could be a bit lucky here, but self-testing is showing a measurable improvement as well.
Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 9:09 pm

Re: Testing Stockfish 15-03-13. 480 Games.

Post by Tomcass »

TESTING STOCKFISH 240313: 480 games

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 4 min +0 sec/ game
No tablebases

Stockfish 240313 – Criter 1.6 +7/=24/-9 19.0 – 21.0 47.50%
Stockfish 240313 – Deep Rybka 4.1 +13/=23/-4 24.5 – 15.5 61.25%
Stockfish 240313 – Houdini 3.0Pro +7/=23/-10 18.5 – 21.5 46.25%

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 min +2 sec/ game
No tablebases

Stockfish 240313 – Criter 1.6 +9/=28/-3 23.0 – 17.0 57.50%
Stockfish 240313 – Deep Rybka 4.1 +13/=20/-7 23.0 – 17.0 57.50%
Stockfish 240313 – Houdini 3.0Pro +3/=19/-18 12.5 – 27.5.0 31.25%

Overall average with 6 cores : ( 120.5 – 119.5) : 50.21%

X6 240 games: http://www.mediafire.com/?iyta1diprq1tx81

i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12
Time control: 4 min + 0 sec/ game
Ponder: Off
No tablebases

Stockfish 240313 - Critter 1.6a +7/=30/-3 22.0 – 18.0 55.00%
Stockfish 240313 - Deep Rybka +8/=23/-9 19.5 – 20.5 48.75%
Stockfish 240313 - Houdini3.0Pro +5/=12/-23 11.0 – 29.0 27.50%

i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12

Time control: 2 min + 2 sec/ game
Ponder: Off
No tablebases

Stockfish 240313 - Critter 1.6a +5/=25/-5 20.0 – 20.0 50.00%
Stockfish 240313 - Deep Rybka +9/=22/-9 20.0 – 20.0 50.00%
Stockfish 240313 - Houdini3.0 Pro +2/=26/-12 15.0 – 25.0 37.50%

Overall average with 4 cores ( 107.50 – 132.50) : 44.79%

X4 240 games: http://www.mediafire.com/?l0iwddag58kn14k

… and after 480 games: 243.5 – 236.5 : 47.50%.

Against Critter 1.6: 52.50%, against Deep Rybka4: 54.37% and against Houdini 3.0 Pro: 35.62%

Regards,

Tom.
User avatar
geots
Posts: 4790
Joined: Sat Mar 11, 2006 12:42 am

For Tom- or Maybe Anyone Who Knows!!

Post by geots »

Is there a place I can go to pick out maybe this version or a few others to download. Please don't tell me we have private beta stockfishes. I would really like to see what this one and a few more will do in comparison tests with 2.3.1 ag. same opponents on 1 core.

I have a couple 4 core i5s, as you know, and am ordering an i7 6-core system when I decide on the best one where hyperthr. can be disabled. Because they don't make them without it. And I would NEVER run one game with it enabled.

I want to get 1-core results because I know the great majority of users just don't have 6-core systems- and I think they would like to see what the versions will do for them. Plus you know I have always admired your work- but I am not sold on the idea that 2+2 is a suitable control for 6-cores.

At any rate, are these private or can I download a few?



Best,

george
Kohflote
Posts: 219
Joined: Wed Sep 19, 2007 11:07 am
Location: Singapore

Re: For Tom- or Maybe Anyone Who Knows!!

Post by Kohflote »

Hi George,

Different versions of Stockfish can be downloaded from here:

http://abrok.eu/stockfish/

Best wishes,
Koh, Kah Huat
User avatar
geots
Posts: 4790
Joined: Sat Mar 11, 2006 12:42 am

Re: For Tom- or Maybe Anyone Who Knows!!

Post by geots »

Kohflote wrote:Hi George,

Different versions of Stockfish can be downloaded from here:

http://abrok.eu/stockfish/

Best wishes,
Koh, Kah Huat



Thank you very much- I shall have a look!


Best to you,

george
Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 9:09 pm

Re: Testing Stockfish 15-03-13. 480 Games.

Post by Tomcass »

TESTING IPMAN FIRST COMPILE: 480 GAMES STOCKFISH 010413

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 4 min +0 sec/ game
No tablebases

Stockfish 010413 – Criter 1.6 +9/=27/-4 22.5 – 17.5 56.25%
Stockfish 010413 – Deep Rybka 4.1 +15/=19/-6 24.5 – 15.5 61.25%
Stockfish 010413 – Houdini 3.0Pro +6/=18/-16 15.0 – 25.0 37.50%

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 min +2 sec/ game
No tablebases

Stockfish 010413 – Criter 1.6 +11/=23/-6 22.5 – 17.5 56.25%
Stockfish 010413 – Deep Rybka 4.1 +13/=20/-7 23.0 – 17.0 57.50%
Stockfish 010413 – Houdini 3.0Pro +1/=24/-15 13.0 – 27.0 32.50%

Overall average with 6 cores : ( 120.5 – 119.5) : 50.21%
X6 240 games http://www.mediafire.com/?7aq8hvq3qofbiqp

i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12
Time control: 4 min + 0 sec/ game
Ponder: Off
No tablebases

Stockfish 010413 - Critter 1.6a +6/=26/-8 19.0 – 21.0 47.50%
Stockfish 010413 - Deep Rybka +19/=14/-7 26.0 – 14.0 65.00%
Stockfish 010413 - Houdini3.0Pro +5/=18/-17 14.0 – 26.0 35.00%

i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12
Time control: 2 min + 2 sec/ game
Ponder: Off
No tablebases

Stockfish 010413 - Critter 1.6a +9/=20/-11 19.0 – 21.0 47.50%
Stockfish 010413 - Deep Rybka +13/=23/-4 24.5 – 15.5 61.25%
Stockfish 010413 - Houdini3.0 Pro +4/=26/-10 17.0 – 23.0 42.50%

Overall average with 4 cores ( 119.50 – 120.50) : 49.79%
X4 240 games http://www.mediafire.com/?8ic3im34y3l8a91

Global Score: 50.00%

Against
Critter: 51.87% Deep Rybka: 61.25% Houdini 3.0 Pro: 36.87%

This is a very good compile. Good try, IPMAN!.

Regards,

Tom
Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 9:09 pm

Re: Testing Stockfish 15-03-13. 480 Games.

Post by Tomcass »

TESTING IPMAN SECOND COMPILE: 480 GAMES STOCKFISH 050413

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 4 min +0 sec/ game
No tablebases

Stockfish 050413 – Criter 1.6 +3/=30/-7 18.0 – 22.0 45.00%
Stockfish 050413 – Deep Rybka 4.1 +11/=20/-9 21.0 – 19.0 52.50%
Stockfish 050413 – Houdini 3.0Pro +4/=19/-17 13.5 – 26.5 33.75%

I7 980 3.33 Ghz.
6 real cores
Ponder: Off.
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 min +2 sec/ game
No tablebases

Stockfish 050413 – Criter 1.6 +7/=29/-4 21.5 – 18.5 53.75%
Stockfish 050413 – Deep Rybka 4.1 +8/=17/-15 16.5 – 23.5 41.25%
Stockfish 050413 – Houdini 3.0Pro +2/=18/-20 11.0 – 29.0 27.50%

Overall average with 6 cores : ( 101.5 – 138.5) : 42.29%

X6 240 games http://www.mediafire.com/?oj44x15fx9ihn5u


i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12
Time control: 4 min + 0 sec/ game
Ponder: Off
No tablebases

Stockfish 050413 - Critter 1.6a +13/=21/-6 23.5 – 16.5 58.75%
Stockfish 050413 - Deep Rybka +10/=23/-7 21.5 – 18.5 53.75%
Stockfish 050413 - Houdini3.0Pro +3/=25/-12 15.5 – 24.5 38.75%

i7 975 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Fritz 12
Time control: 2 min + 2 sec/ game
Ponder: Off
No tablebases

Stockfish 050413 - Critter 1.6a +5/=31/-4 20.5 – 19.5 51.25%
Stockfish 050413 - Deep Rybka +10/=26/-4 23.0 – 17.0 57.50%
Stockfish 050413 - Houdini3.0 Pro +5/=23/-12 16.5 – 23.5 41.25%

Overall average with 4 cores ( 120.5 – 119-5) : 50.21%

X4 240 games http://www.mediafire.com/?ofa2nmpvwccjwsk

Global Score: 46.25%

Against
Critter: 52.19% Deep Rybka: 51.25% Houdini 3.0 Pro: 35.31%


Regards,

Tom.
Jouni
Posts: 3291
Joined: Wed Mar 08, 2006 8:15 pm

Re: Testing Stockfish 15-03-13. 480 Games.

Post by Jouni »

There has been so many +3...+5 elo patches now, that I wonder why Houdini level hasn't reached yet with SF :?: .
Jouni
User avatar
Eelco de Groot
Posts: 4567
Joined: Sun Mar 12, 2006 2:40 am
Full name:   

Re: Testing Stockfish 15-03-13. 480 Games.

Post by Eelco de Groot »

Jouni wrote:There has been so many +3...+5 elo patches now, that I wonder why Houdini level hasn't reached yet with SF :?: .
The gap with Houdini is very large. 50%-35%=15, that is about 15*7 =105 elo. And selftesting will always seem increase the elo more. The best patches have to be retained and for that you need also to do regression testing of all the changes against the last stable Stockfish because otherwise nonfunctional patches might turn out to be not so nonfunctional as thought.. If Stockfish would really pass this last version of Critter 1.6 it would already be something, that is still by no means certain now, and a real achievement :) I think Richard Vida could pass Stockfish again if he wanted, but that is another matter...

Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan