SPCC: Testrun of Stockfish 13 finished

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
pohl4711
Posts: 1502
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

SPCC: Testrun of Stockfish 13 finished

Post by pohl4711 » Mon Feb 22, 2021 5:42 pm

AB-testrun of Stockfish 13 avx2 finished. +39 Elo to Stockfish 12.

https://www.sp-cc.de

Remarkabale is the head-to-head vs. Fat Fritz 2. Fat Fritz 2 won it. I did not expect that.

Code: Select all

4 Stockfish 13 210218      : 3723 8000 (+3992,=3890,-118), 74.2 %

SF Fat Fritz 2 avx2        : 1000 (+ 59,=872,- 69), 49.5 %
RubiChess 2.0 avx2         : 1000 (+784,=216,-  0), 89.2 %
Slow Chess 2.5 avx2        : 1000 (+701,=297,-  2), 85.0 %
KomodoDragon 1.0 avx2      : 1000 (+208,=769,- 23), 59.3 %
Ethereal 12.75 avx2        : 1000 (+744,=256,-  0), 87.2 %
Nemorino 6.00 avx2         : 1000 (+636,=361,-  3), 81.7 %
Stockfish 12 200902        : 1000 (+125,=857,- 18), 55.4 %
Houdini 6 pext             : 1000 (+735,=262,-  3), 86.6 %

(Perhaps you have to clear your browsercache or reload the website)

Modern Times
Posts: 2711
Joined: Thu Jun 07, 2012 9:02 pm

Re: SPCC: Testrun of Stockfish 13 finished

Post by Modern Times » Tue Feb 23, 2021 1:18 am

pohl4711 wrote:
Mon Feb 22, 2021 5:42 pm

Remarkabale is the head-to-head vs. Fat Fritz 2. Fat Fritz 2 won it. I did not expect that.
Chessbase will no doubt be happy ! This has been the basis for their "New #1" marketing campaign, that FF2 is stronger head-to-head with Stockfish which is all that matters to them. 50.5% is the narrowest of margins though.

Like many people I can't replicate results on their website. I'm currently running a match between the two with your 5mvs_balanced pgn, 2871 positions with reversed sides. 90s + 1s. That match currently stands at:

Code: Select all

Score of Stockfish 13 64-bit vs Fat Fritz 2 64-bit: 375 - 199 - 4712 [0.517]
Elo difference: 11.6 +/- 3.1, LOS: 100.0 %, DrawRatio: 89.1 %

5286 of 5742 games finished.
.

Opinions expressed here are my own, and not necessarily those of the CCRL Group.

ernest
Posts: 1928
Joined: Wed Mar 08, 2006 7:30 pm

Re: SPCC: Testrun of Stockfish 13 finished

Post by ernest » Tue Feb 23, 2021 1:26 am

pohl4711 wrote:
Mon Feb 22, 2021 5:42 pm

Remarkabale is the head-to-head vs. Fat Fritz 2. Fat Fritz 2 won it. I did not expect that.

Code: Select all

4 Stockfish 13 210218      : 3723 8000 (+3992,=3890,-118), 74.2 %

SF Fat Fritz 2 avx2        : 1000 (+ 59,=872,- 69), 49.5 % 
Thanks again for all your work, Stefan !
49.5 %
...but 95 % error is a little over 1 %

Still consistent with your ratings...

pohl4711
Posts: 1502
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: SPCC: Testrun of Stockfish 13 finished

Post by pohl4711 » Tue Feb 23, 2021 3:25 am

ernest wrote:
Tue Feb 23, 2021 1:26 am
pohl4711 wrote:
Mon Feb 22, 2021 5:42 pm

Remarkabale is the head-to-head vs. Fat Fritz 2. Fat Fritz 2 won it. I did not expect that.

Code: Select all

4 Stockfish 13 210218      : 3723 8000 (+3992,=3890,-118), 74.2 %

SF Fat Fritz 2 avx2        : 1000 (+ 59,=872,- 69), 49.5 % 
Thanks again for all your work, Stefan !
49.5 %
...but 95 % error is a little over 1 %

Still consistent with your ratings...
Of course. 50.5% in 1000 games is clearly inside errorbar. And Fat Fritz 2 overall rating is still below SF 13.
In my ratinglist, I cannot play 20000 games or so in one testrun...too much effort.

Post Reply