Status Quo 2016-1 , Stockfish - Komodo
Moderators: hgm, Rebel, chrisw
-
- Posts: 3293
- Joined: Wed Mar 08, 2006 8:15 pm
Re: Status Quo 2016-1 , Stockfish - Komodo
There are not much difference in nps, depth or evaluation between engines, so no wonder 50% score now . But why Komodo displays 66,1% hash usage whole time?
Jouni
-
- Posts: 184
- Joined: Sun Nov 01, 2015 3:15 am
Re: Status Quo 2016-1 , Stockfish - Komodo
Very nice setup, I'm jealous. Except for the hash, this is basically a replay of the TCEC Season 8 Super Final.
-
- Posts: 1494
- Joined: Thu Mar 30, 2006 2:08 pm
Re: Status Quo 2016-1 , Stockfish - Komodo
Komodo only resets the hashfull count when the hash is cleared or resized. This is how Don programmed it, although I know most programs reset it for each new search.Jouni wrote:There are not much difference in nps, depth or evaluation between engines, so no wonder 50% score now . But why Komodo displays 66,1% hash usage whole time?
-
- Posts: 3293
- Joined: Wed Mar 08, 2006 8:15 pm
-
- Posts: 41466
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
Re: Status Quo 2016-1 , Stockfish - Komodo
Brilliant!Hugo wrote:Hi engine lovers, and a happy new year 2016
I have started an engine macht between Komodo and Stockfish.
Engines use 24 cpu, no hyperthreading, ponder OFF
8 GB of Hashtables
Syzyggy 3-4-5-6
100 games, 60 + 0 timer
using a set of 50 startpossitions 2- 8 moves
Komodo contemp = 0
Will post later games and crosstable
for intrested observers you can watch the games with 15 sec update frequency:
http://clemens-keck.de/screen/
regards, Clemens Keck
Looking forward to some high quality games.
Thanks Clemens.
gbanksnz at gmail.com
-
- Posts: 1971
- Joined: Wed Jul 13, 2011 9:04 pm
- Location: Madrid, Spain.
Re: Status Quo 2016-1, Stockfish - Komodo.
Hello:
Regarding the asymmetric confidence intervals of Fritz GUI (I suppose it is Fritz GUI), I managed to get most of the numbers. At this moment:
I reached the following equations for the score bounds after many trials. I hope no typos:
And then convert these scores into Elo differences with the well-known 400·log10[score/(1 - score)]. Here is what I obtained with Excel after rounding to the nearest integers:
The only difference is in the upper Elo difference of 3-sigma (+191 versus +192). The bias of the centre of each interval is z·ln(2·µ) above the score.
I do not claim it is the definitive formula for Fritz GUI. Please see:
Stockfish 070114 Vs Stockfish 070114 8 logical cores vs 4
There I replace µ + z·ln(2·µ) by 0.5 + (z/3)·ln(2·µ) and I get somewhat similar 3-sigma confidence intervals but not so similar than Clemens' current match with the other formula.
Is there any known distribution with my first formula or is it a lucky good fit for this sample? More samples to test the formula are needed, of course.
Regards from Spain.
Ajedrecista.
Excellent! Thanks for sharing.Hugo wrote:Hi engine lovers, and a happy new year 2016
I have started an engine macht between Komodo and Stockfish.
Engines use 24 cpu, no hyperthreading, ponder OFF
8 GB of Hashtables
Syzyggy 3-4-5-6
100 games, 60 + 0 timer
using a set of 50 startpossitions 2- 8 moves
Komodo contemp = 0
Will post later games and crosstable
for intrested observers you can watch the games with 15 sec update frequency:
http://clemens-keck.de/screen/
regards, Clemens Keck
Regarding the asymmetric confidence intervals of Fritz GUI (I suppose it is Fritz GUI), I managed to get most of the numbers. At this moment:
Code: Select all
Komodo 9.3 64-bit x24 - Stockfish 030116 64 POPCNT; 39 Games, +5 =30 -4, 51.3%, TP=+9 Elo, 68%->[-11,+65],
95%->[-31,+124], 99.7%->[-51,+191], predicted moves=63.6%
Code: Select all
µ = (wins + 0.5*draws)/games
Assuming that Fritz GUI reports 1-sigma, 2-sigma and 3-sigma confidence intervals, for z-sigma confidence interval:
(Lower score bound) = µ + z·ln(2·µ) - (2·z/3)·sqrt[µ·(1 - µ)/(games - 1)]
(Upper score bound) = µ + z·ln(2·µ) + (2·z/3)·sqrt[µ·(1 - µ)/(games - 1)]
Code: Select all
1-sigma: [-11, +65]
2-sigma: [-31, +124]
3-sigma: [-51, +192]
I do not claim it is the definitive formula for Fritz GUI. Please see:
Stockfish 070114 Vs Stockfish 070114 8 logical cores vs 4
There I replace µ + z·ln(2·µ) by 0.5 + (z/3)·ln(2·µ) and I get somewhat similar 3-sigma confidence intervals but not so similar than Clemens' current match with the other formula.
Is there any known distribution with my first formula or is it a lucky good fit for this sample? More samples to test the formula are needed, of course.
Regards from Spain.
Ajedrecista.
-
- Posts: 184
- Joined: Sun Nov 01, 2015 3:15 am
Re: Status Quo 2016-1 , Stockfish - Komodo
What are the rules for adjudicating these games?
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: Status Quo 2016-1 , Stockfish - Komodo
If you only show a "peak" value you would expect it to reach some "peak" and stick there. I have a "data" percentage for the parallel search data used in Crafty. I keep a peak value for that that displays the "peak usage" after each search. Not average or whatever, as I want to know if it ever reaches 100% which would indicate a need for a change in the size.Jouni wrote:Yes I know. But why it's exactly 66,1%?
-
- Posts: 782
- Joined: Tue Dec 01, 2009 11:10 am
Re: Status Quo 2016-1 , Stockfish - Komodo
huh, I must say I dont know this. The Fritz 15 gui is in basic settings.Dan Cooper wrote:What are the rules for adjudicating these games?
I think draw comes from move repetition or 50 move rule?
C.K.
-
- Posts: 69
- Joined: Wed Mar 08, 2006 8:19 pm
Re: Status Quo 2016-1 , Stockfish - Komodo
The Fritz 15 GUI adjudication options are located in the File-Options-Game tab. If you read the help file located at :
http://help.chessbase.com/Fritz/15/Eng/ ... 000065.htm
you can see some of the specifics. Unfortunately I do not see any details on the Draw option.
http://help.chessbase.com/Fritz/15/Eng/ ... 000065.htm
you can see some of the specifics. Unfortunately I do not see any details on the Draw option.