Komodo 9.3 vs Stockfish

Discussion of computer chess matches and engine tournaments.

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
beram
Posts: 1187
Joined: Wed Jan 06, 2010 2:11 pm

Re: Komodo 9.3 vs Stockfish

Post by beram » Fri Dec 25, 2015 9:12 am

beram wrote:
beram wrote:last match at 15m10s on my AMD 1090T 4cores each
57% (+10-3=37) victory for latest SF141215 against Komodo 9.3(!)

Code: Select all

SF 141215 - Komodo 9.3, Rapid 15m+10s
                             
1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
..
And again a victory for SF181215, this time 54% (+8-4=38)
With a double win from both sides in a Kings Indian testposition as well as in the Kings gambit (1e4 - e5 2f4 -exf4)

Code: Select all

SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50
And for Stockfish 231215 64 +10/=36/-4 56.00% 28.0/50
with again a double win from both sides of the board. This time in a Dutch gambit and again in the Kings Indian

Code: Select all

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50

Jouni
Posts: 2223
Joined: Wed Mar 08, 2006 7:15 pm

Re: Komodo 9.3 vs Stockfish

Post by Jouni » Fri Dec 25, 2015 6:23 pm

In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
Jouni

beram
Posts: 1187
Joined: Wed Jan 06, 2010 2:11 pm

Re: Komodo 9.3 vs Stockfish

Post by beram » Sat Dec 26, 2015 11:24 am

Jouni wrote:In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
hi Jouni,
I have tested Komodo with contempt=0 for long time but besides more draws Komodo didn't do better against the SF dev from sept-nov 15
Therefore i have stopped testing with that setting

So I conclude that under these TC conditions 15m10s, 4cores on my AMD 1090T, kind of intermediate between 4/40 and 40/40 CCRL level, the latest SF dev’s from december, are clearly stronger than Komodo 9.3 in head to head matches
In 250 games the score is 136 – 112 (+47 – 25 = 178) for SFdev (from dec) and that is 54,4 % at these Rapid TC conditions, ~50min avg game duration with 25-26 ply depth in early middlegame
What is remarkable and a show of its strenght, is that twelve times in these last 250 games, SF managed to win positions from both sides of the board.
These where not only open positions as for Benko gambit(2x), Kings gambit, Dutch gambit, but also two different closed Kings Indian positions in which SF also feels like a fish in the water against the second strongest engine in the world
When the Stockfish authors would bring out a new Stockfish 7 now, with some speedup because of compile optimalization it would top all of the most common lists CEGT CCRL

Code: Select all

SF 021215 - Komodo 9.3, Blitz 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
1   Komodo 9.3 64-bit    +8/=34/-8 50.00%   25.0/50  0.00
2   Stockfish 021215 64  +8/=34/-8 50.00%   25.0/50  0.00

SF 051215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 051215 64   +35  +11/=33/-6 55.00%   27.5/50
2   Komodo 9.3 64-bit     -35  +6/=33/-11 45.00%   22.5/50
SF 141215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50 
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50

APassionForCriminalJustic
Posts: 415
Joined: Sat May 24, 2014 7:16 am

Re: Komodo 9.3 vs Stockfish

Post by APassionForCriminalJustic » Sat Dec 26, 2015 8:58 pm

beram wrote:
Jouni wrote:In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
hi Jouni,
I have tested Komodo with contempt=0 for long time but besides more draws Komodo didn't do better against the SF dev from sept-nov 15
Therefore i have stopped testing with that setting

So I conclude that under these TC conditions 15m10s, 4cores on my AMD 1090T, kind of intermediate between 4/40 and 40/40 CCRL level, the latest SF dev’s from december, are clearly stronger than Komodo 9.3 in head to head matches
In 250 games the score is 136 – 112 (+47 – 25 = 178) for SFdev (from dec) and that is 54,4 % at these Rapid TC conditions, ~50min avg game duration with 25-26 ply depth in early middlegame
What is remarkable and a show of its strenght, is that twelve times in these last 250 games, SF managed to win positions from both sides of the board.
These where not only open positions as for Benko gambit(2x), Kings gambit, Dutch gambit, but also two different closed Kings Indian positions in which SF also feels like a fish in the water against the second strongest engine in the world
When the Stockfish authors would bring out a new Stockfish 7 now, with some speedup because of compile optimalization it would top all of the most common lists CEGT CCRL

Code: Select all

SF 021215 - Komodo 9.3, Blitz 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
1   Komodo 9.3 64-bit    +8/=34/-8 50.00%   25.0/50  0.00
2   Stockfish 021215 64  +8/=34/-8 50.00%   25.0/50  0.00

SF 051215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 051215 64   +35  +11/=33/-6 55.00%   27.5/50
2   Komodo 9.3 64-bit     -35  +6/=33/-11 45.00%   22.5/50
SF 141215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50 
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50
I'm not buying your results... your AMD hardware certainly doesn't help anything. Komodo is not the second strongest in the world... You've played not nearly enough games... nor is your time control sufficient. Stockfish loves crappy hardware... try TCEC conditions. Stockfish didn't look so strong there did it? Right now we have two top engines in the world depending on the CONDITIONS and hardware those of which are played. Those are the facts.

beram
Posts: 1187
Joined: Wed Jan 06, 2010 2:11 pm

Re: Komodo 9.3 vs Stockfish

Post by beram » Sat Dec 26, 2015 10:39 pm

APassionForCriminalJustic wrote:
beram wrote:
Jouni wrote:In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
hi Jouni,
I have tested Komodo with contempt=0 for long time but besides more draws Komodo didn't do better against the SF dev from sept-nov 15
Therefore i have stopped testing with that setting

So I conclude that under these TC conditions 15m10s, 4cores on my AMD 1090T, kind of intermediate between 4/40 and 40/40 CCRL level, the latest SF dev’s from december, are clearly stronger than Komodo 9.3 in head to head matches
In 250 games the score is 136 – 112 (+47 – 25 = 178) for SFdev (from dec) and that is 54,4 % at these Rapid TC conditions, ~50min avg game duration with 25-26 ply depth in early middlegame
What is remarkable and a show of its strenght, is that twelve times in these last 250 games, SF managed to win positions from both sides of the board.
These where not only open positions as for Benko gambit(2x), Kings gambit, Dutch gambit, but also two different closed Kings Indian positions in which SF also feels like a fish in the water against the second strongest engine in the world
When the Stockfish authors would bring out a new Stockfish 7 now, with some speedup because of compile optimalization it would top all of the most common lists CEGT CCRL

Code: Select all

SF 021215 - Komodo 9.3, Blitz 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
1   Komodo 9.3 64-bit    +8/=34/-8 50.00%   25.0/50  0.00
2   Stockfish 021215 64  +8/=34/-8 50.00%   25.0/50  0.00

SF 051215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 051215 64   +35  +11/=33/-6 55.00%   27.5/50
2   Komodo 9.3 64-bit     -35  +6/=33/-11 45.00%   22.5/50
SF 141215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50 
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50
I'm not buying your results... your AMD hardware certainly doesn't help anything. Komodo is not the second strongest in the world... You've played not nearly enough games... nor is your time control sufficient. Stockfish loves crappy hardware... try TCEC conditions. Stockfish didn't look so strong there did it? Right now we have two top engines in the world depending on the CONDITIONS and hardware those of which are played. Those are the facts.
You are good at insulting but just stay with the facts
ply depth 30 or 34 on 16 cores and still not seeing any danger
http://www.talkchess.com/forum/viewtopi ... 51&t=58692

JJJ
Posts: 1317
Joined: Sat Apr 19, 2014 11:47 am

Re: Komodo 9.3 vs Stockfish

Post by JJJ » Sat Dec 26, 2015 11:34 pm

Stockfish dev almost always beat Komodo on other hardware that TCEC hardware, but in the End, Komodo is TCEC champion with a good margin.


So for now, your test are just proving nothing new to me. And you re calling stockfish number 1 if you want, but for me it is not a fair statment, just your preference since the begining.


If Komodo team is now working on winning with short time control too, it could be also the end of Stockfish domination on short time control and maybe on 4-8 core machines.

stavros
Posts: 160
Joined: Tue Dec 02, 2014 12:29 am

Re: Komodo 9.3 vs Stockfish

Post by stavros » Sun Dec 27, 2015 9:29 pm

beram wrote:
APassionForCriminalJustic wrote:
beram wrote:
Jouni wrote:In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
hi Jouni,
I have tested Komodo with contempt=0 for long time but besides more draws Komodo didn't do better against the SF dev from sept-nov 15
Therefore i have stopped testing with that setting

So I conclude that under these TC conditions 15m10s, 4cores on my AMD 1090T, kind of intermediate between 4/40 and 40/40 CCRL level, the latest SF dev’s from december, are clearly stronger than Komodo 9.3 in head to head matches
In 250 games the score is 136 – 112 (+47 – 25 = 178) for SFdev (from dec) and that is 54,4 % at these Rapid TC conditions, ~50min avg game duration with 25-26 ply depth in early middlegame
What is remarkable and a show of its strenght, is that twelve times in these last 250 games, SF managed to win positions from both sides of the board.
These where not only open positions as for Benko gambit(2x), Kings gambit, Dutch gambit, but also two different closed Kings Indian positions in which SF also feels like a fish in the water against the second strongest engine in the world
When the Stockfish authors would bring out a new Stockfish 7 now, with some speedup because of compile optimalization it would top all of the most common lists CEGT CCRL

Code: Select all

SF 021215 - Komodo 9.3, Blitz 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
1   Komodo 9.3 64-bit    +8/=34/-8 50.00%   25.0/50  0.00
2   Stockfish 021215 64  +8/=34/-8 50.00%   25.0/50  0.00

SF 051215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 051215 64   +35  +11/=33/-6 55.00%   27.5/50
2   Komodo 9.3 64-bit     -35  +6/=33/-11 45.00%   22.5/50
SF 141215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50 
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50
I'm not buying your results... your AMD hardware certainly doesn't help anything. Komodo is not the second strongest in the world... You've played not nearly enough games... nor is your time control sufficient. Stockfish loves crappy hardware... try TCEC conditions. Stockfish didn't look so strong there did it? Right now we have two top engines in the world depending on the CONDITIONS and hardware those of which are played. Those are the facts.
You are good at insulting but just stay with the facts
ply depth 30 or 34 on 16 cores and still not seeing any danger
http://www.talkchess.com/forum/viewtopi ... 51&t=58692
interesting results! let them defend desperately komodo, 99% of users know the truth:)
as for adam and jean baptiste have lost their reliablity insulting to anyone who dispute komodo.
oh and u have to buy a 2048 core pc and use VLTC 480+120 and 1200 games , i didnt know that i should have a monster pc to buy komodo 9.3 :)

APassionForCriminalJustic
Posts: 415
Joined: Sat May 24, 2014 7:16 am

Re: Komodo 9.3 vs Stockfish

Post by APassionForCriminalJustic » Sun Dec 27, 2015 9:57 pm

beram wrote:
APassionForCriminalJustic wrote:
beram wrote:
Jouni wrote:In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
hi Jouni,
I have tested Komodo with contempt=0 for long time but besides more draws Komodo didn't do better against the SF dev from sept-nov 15
Therefore i have stopped testing with that setting

So I conclude that under these TC conditions 15m10s, 4cores on my AMD 1090T, kind of intermediate between 4/40 and 40/40 CCRL level, the latest SF dev’s from december, are clearly stronger than Komodo 9.3 in head to head matches
In 250 games the score is 136 – 112 (+47 – 25 = 178) for SFdev (from dec) and that is 54,4 % at these Rapid TC conditions, ~50min avg game duration with 25-26 ply depth in early middlegame
What is remarkable and a show of its strenght, is that twelve times in these last 250 games, SF managed to win positions from both sides of the board.
These where not only open positions as for Benko gambit(2x), Kings gambit, Dutch gambit, but also two different closed Kings Indian positions in which SF also feels like a fish in the water against the second strongest engine in the world
When the Stockfish authors would bring out a new Stockfish 7 now, with some speedup because of compile optimalization it would top all of the most common lists CEGT CCRL

Code: Select all

SF 021215 - Komodo 9.3, Blitz 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
1   Komodo 9.3 64-bit    +8/=34/-8 50.00%   25.0/50  0.00
2   Stockfish 021215 64  +8/=34/-8 50.00%   25.0/50  0.00

SF 051215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 051215 64   +35  +11/=33/-6 55.00%   27.5/50
2   Komodo 9.3 64-bit     -35  +6/=33/-11 45.00%   22.5/50
SF 141215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50 
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50
I'm not buying your results... your AMD hardware certainly doesn't help anything. Komodo is not the second strongest in the world... You've played not nearly enough games... nor is your time control sufficient. Stockfish loves crappy hardware... try TCEC conditions. Stockfish didn't look so strong there did it? Right now we have two top engines in the world depending on the CONDITIONS and hardware those of which are played. Those are the facts.
You are good at insulting but just stay with the facts
ply depth 30 or 34 on 16 cores and still not seeing any danger
http://www.talkchess.com/forum/viewtopi ... 51&t=58692
What facts?! With your hardware?! Right.... how about you update that junk?!

APassionForCriminalJustic
Posts: 415
Joined: Sat May 24, 2014 7:16 am

Re: Komodo 9.3 vs Stockfish

Post by APassionForCriminalJustic » Sun Dec 27, 2015 9:58 pm

stavros wrote:
beram wrote:
APassionForCriminalJustic wrote:
beram wrote:
Jouni wrote:In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
hi Jouni,
I have tested Komodo with contempt=0 for long time but besides more draws Komodo didn't do better against the SF dev from sept-nov 15
Therefore i have stopped testing with that setting

So I conclude that under these TC conditions 15m10s, 4cores on my AMD 1090T, kind of intermediate between 4/40 and 40/40 CCRL level, the latest SF dev’s from december, are clearly stronger than Komodo 9.3 in head to head matches
In 250 games the score is 136 – 112 (+47 – 25 = 178) for SFdev (from dec) and that is 54,4 % at these Rapid TC conditions, ~50min avg game duration with 25-26 ply depth in early middlegame
What is remarkable and a show of its strenght, is that twelve times in these last 250 games, SF managed to win positions from both sides of the board.
These where not only open positions as for Benko gambit(2x), Kings gambit, Dutch gambit, but also two different closed Kings Indian positions in which SF also feels like a fish in the water against the second strongest engine in the world
When the Stockfish authors would bring out a new Stockfish 7 now, with some speedup because of compile optimalization it would top all of the most common lists CEGT CCRL

Code: Select all

SF 021215 - Komodo 9.3, Blitz 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
1   Komodo 9.3 64-bit    +8/=34/-8 50.00%   25.0/50  0.00
2   Stockfish 021215 64  +8/=34/-8 50.00%   25.0/50  0.00

SF 051215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 051215 64   +35  +11/=33/-6 55.00%   27.5/50
2   Komodo 9.3 64-bit     -35  +6/=33/-11 45.00%   22.5/50
SF 141215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50 
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50
I'm not buying your results... your AMD hardware certainly doesn't help anything. Komodo is not the second strongest in the world... You've played not nearly enough games... nor is your time control sufficient. Stockfish loves crappy hardware... try TCEC conditions. Stockfish didn't look so strong there did it? Right now we have two top engines in the world depending on the CONDITIONS and hardware those of which are played. Those are the facts.
You are good at insulting but just stay with the facts
ply depth 30 or 34 on 16 cores and still not seeing any danger
http://www.talkchess.com/forum/viewtopi ... 51&t=58692
interesting results! let them defend desperately komodo, 99% of users know the truth:)
as for adam and jean baptiste have lost their reliablity insulting to anyone who dispute komodo.
oh and u have to buy a 2048 core pc and use VLTC 480+120 and 1200 games , i didnt know that i should have a monster pc to buy komodo 9.3 :)
We don't need 2048 cores... just not complete garbage... Komodo isn't even developed with AMD.

APassionForCriminalJustic
Posts: 415
Joined: Sat May 24, 2014 7:16 am

Re: Komodo 9.3 vs Stockfish

Post by APassionForCriminalJustic » Sun Dec 27, 2015 10:04 pm

beram wrote:
APassionForCriminalJustic wrote:
beram wrote:
Jouni wrote:In Core i5 Computer Komodo 9.2 has got so far always 50% or over in my test with medium time control vs Stockfish. Settings of course:
King Safety=65
Selectivity=131
Contempt=0
hi Jouni,
I have tested Komodo with contempt=0 for long time but besides more draws Komodo didn't do better against the SF dev from sept-nov 15
Therefore i have stopped testing with that setting

So I conclude that under these TC conditions 15m10s, 4cores on my AMD 1090T, kind of intermediate between 4/40 and 40/40 CCRL level, the latest SF dev’s from december, are clearly stronger than Komodo 9.3 in head to head matches
In 250 games the score is 136 – 112 (+47 – 25 = 178) for SFdev (from dec) and that is 54,4 % at these Rapid TC conditions, ~50min avg game duration with 25-26 ply depth in early middlegame
What is remarkable and a show of its strenght, is that twelve times in these last 250 games, SF managed to win positions from both sides of the board.
These where not only open positions as for Benko gambit(2x), Kings gambit, Dutch gambit, but also two different closed Kings Indian positions in which SF also feels like a fish in the water against the second strongest engine in the world
When the Stockfish authors would bring out a new Stockfish 7 now, with some speedup because of compile optimalization it would top all of the most common lists CEGT CCRL

Code: Select all

SF 021215 - Komodo 9.3, Blitz 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
1   Komodo 9.3 64-bit    +8/=34/-8 50.00%   25.0/50  0.00
2   Stockfish 021215 64  +8/=34/-8 50.00%   25.0/50  0.00

SF 051215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 051215 64   +35  +11/=33/-6 55.00%   27.5/50
2   Komodo 9.3 64-bit     -35  +6/=33/-11 45.00%   22.5/50
SF 141215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)

1   Stockfish 141215 64   +49  +10/=37/-3 57.00%   28.5/50 
2   Komodo 9.3 64-bit     -49  +3/=37/-10 43.00%   21.5/50
SF 181215 - Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 181215 64   +28  +8/=38/-4 54.00%   27.0/50
2   Komodo 9.3 64-bit     -28  +4/=38/-8 46.00%   23.0/50

SF 231215 Komodo 9.3, Rapid 15m+10s  AMD 1090T @3200Mhz, 4cpu (Fritzmark 15,6)
                               
1   Stockfish 231215 64   +42  +10/=36/-4 56.00%   28.0/50
2   Komodo 9.3 64-bit     -42  +4/=36/-10 44.00%   22.0/50
I'm not buying your results... your AMD hardware certainly doesn't help anything. Komodo is not the second strongest in the world... You've played not nearly enough games... nor is your time control sufficient. Stockfish loves crappy hardware... try TCEC conditions. Stockfish didn't look so strong there did it? Right now we have two top engines in the world depending on the CONDITIONS and hardware those of which are played. Those are the facts.
You are good at insulting but just stay with the facts
ply depth 30 or 34 on 16 cores and still not seeing any danger
http://www.talkchess.com/forum/viewtopi ... 51&t=58692
What the results with that junk Opteron PC? 16 crappy cores... small hash... only 5 men tbs... even the time control is well short of the 180 mins at TCEC.

I may come across as insulting... but it is actually that I don't just believe results on mediocre hardware to take as truth...

Post Reply