Komodo 8CPU v Stockfish 8CPU Match

[arunsoorya1309]

Komodo should bounce back. Mark has said he found some bug with time management which is playing in current tournament, so may be next tournament will be more competitive

[shrapnel]

Time Management, my *** !
He will have to fix a lot more than just Time Management, if he wants to get Komodo competitive again !

[lantonov]

The development version of Komodo that is playing here (1702.00) is later than the version playing in TCEC rapid (1692.19)

[shrapnel]

The development version of Komodo that is playing here (1702.00) is later than the version playing in TCEC rapid (1692.19)

LOL ! That puts paid to THAT excuse !
Or maybe its just that the glory days of Komodo are over now, now that Houdart is back with a Bang !
It'll be like the old days where Stockfish and Houdini ruled the roost and Komodo was a distant third !

[Jouni]

How can 2 man team compete in long run with Stockfish team - impossible. Additionally Mark and Larry are not youngster anymore. It's miracle they have so strong engine currently. But Houdart is another thing: a magician?

[Modern Times]

How can 2 man team compete in long run with Stockfish team - impossible. Additionally Mark and Larry are not youngster anymore

Well, I'd say how can SF compete with Komodo, when all SF new ideas and techniques are available to Mark and Larry, but all of Komodo remains secret. Larry is also a GM which is an advantage for them too. So for me, it is surprising that SF can keep pace with Komodo.

[APassionForCriminalJustic]

How can 2 man team compete in long run with Stockfish team - impossible. Additionally Mark and Larry are not youngster anymore

This was meant for Ray Banks not Jouni. I apologize for my error.


I get your logic of course but it is not as simple as "copying" ideas or code from Stockfish and placing them into Komodo. What works for Stockfish may not work for Komodo. You have two old men trying to keep up with an entire framework and community. Just like what Marco himself argued - sure people can look at Stockfish's code - but the greatest advantage of all is that Stockfish is completely community driven. Anyone can contribute. I would actually argue that Larry is not as much of an advantage as what a lot of us‎ would argue. He's one man. Stockfish is effectively proving that despite Komodo's "glorious" evaluation function or wealth of knowledge, it is not able to beat Stockfish in match play (wins versus losses). Houdart is the exact same story. Stockfish already took Houdini's number one spot; it ended up leaving Houdart's mouth foul and ultimately complaining that "Stockfish was effectively destroying commercial engines". There is no "comeback" king. One man is no longer going to run the show.

‎It's actually quite funny that both Mark and Larry are just figuring this out NOW that Komodo had not been thinking long enough at TCEC. I would always set time-usage aggressiveness to +9. All I play are VLTC matches so I can see the obvious benefits of allowing the engines to use as much time as feasibly possible.‎ Nevertheless, Komodo STILL won three TCEC superfinals\championships with the default value of zero (if I am not mistaken). What this whole thing sounds like to me is just an excuse. We all know that Stockfish is better. I do not care if it's with one core or 96. Stockfish is running this show - and it's going to give Houdini the beat down next.

Apparently the newest team Komodo excuse is that 20 cores isn't enough anymore for true success. ‎This was basically stated in TCEC chat awhile ago by Komodo's operator.

It's getting old real quick.‎ All we ever get is excuses with team Komodo. For instance, we need more cores; we need more time; our time management isn't correct; Komodo just isn't as good at blitz versus Stockfish; 25 minute games are too "quick"; there are not enough games played to offset statistical noise; etc,. It is really quite pathetic.

[shrapnel]

+1.
Maybe it is time that the Komodo Team retired gracefully from the computer chess scene, while they still have a decent Engine.
As for whether Houdart is up to the Stockfish challenge, only time will tell.

[arunsoorya1309]

I think folks are over reacting ... K just had a few average months and already you folks are talking about retirement of the engine ... its an insult to lary and mark

[voyagerOne]

Komodo team just needs to focus on machine learning for lmr...

Which SF currently does very well at thanks to CMH tables.