Komodo time management

[Damir]

is very bad. Can Komodo team take a look at Stockfish code, and implement it's time management in Komodo ? In a couple of games I have played, Komodo is always in time trouble, with only few seconds on the clock, while Stockfish has more than 1 MIN...

[mjlef]

is very bad. Can Komodo team take a look at Stockfish code, and implement it's time management in Komodo ? In a couple of games I have played, Komodo is always in time trouble, with only few seconds on the clock, while Stockfish has more than 1 MIN...

Well, taking another program's time allocation scheme would be wrong, and the way Komodo works, maybe not even possible. If you can provide some details such as what time controls you are using, if you are playing online and so on, I could suggest some settings that could help.

For online play, lag on the internet will require you to use a higher "Overhead ms" value. At least 100, but if your lag is high, use a larger number. The "Aggressiveness" parameter lets you also control how time is used up throughout the game. Look at the README.htm file for more information.

[hgm]

is very bad. Can Komodo team take a look at Stockfish code, and implement it's time management in Komodo ? In a couple of games I have played, Komodo is always in time trouble, with only few seconds on the clock, while Stockfish has more than 1 MIN...

Do you even have proof that this is bad? What TC is this anyway, incremental or sudden death? It could also be evidence that Stockfish doensn't efficienly use its time.

That being said, in sudden-death games it is usually better not to stick religiously to your own time-usage scheme, but adapt it a bit in the direction of the opponent. I.e. if the opponent initially uses 3x as much time as you think optimal, it would be better to start using 2-2.5x your optimal time as well. You don't want him to outsearch you by a factor 3, even though that will bring him in time trouble later, as later the game will likely already be decided in his advantage. Likewise, if he is faster than you think you should be, prevent that his remaining time ever increases to more than 1.5x yours, by speeding up if necessary.

[syzygy]

is very bad. Can Komodo team take a look at Stockfish code, and implement it's time management in Komodo ? In a couple of games I have played, Komodo is always in time trouble, with only few seconds on the clock, while Stockfish has more than 1 MIN...

I hope that SF at least was winning those games, because if it was not, it should have used that time!!

If a player loses a game without being low on time, he could have used his time better.

[APassionForCriminalJustic]

is very bad. Can Komodo team take a look at Stockfish code, and implement it's time management in Komodo ? In a couple of games I have played, Komodo is always in time trouble, with only few seconds on the clock, while Stockfish has more than 1 MIN...

I have never seen any issues with Komodo’s time usage (I find it to be rather intelligent). I actually have mine set to time usage 9 (the most aggressive setting), and it works very well. I play either 5/5, 60/29, 55/5 or 120/60. Sudden death matches which involve no increments are ludicrous in my opinion, so steer clear of those. But you need to provide your time control used man.

[APassionForCriminalJustic]

is very bad. Can Komodo team take a look at Stockfish code, and implement it's time management in Komodo ? In a couple of games I have played, Komodo is always in time trouble, with only few seconds on the clock, while Stockfish has more than 1 MIN...

Well, taking another program's time allocation scheme would be wrong, and the way Komodo works, maybe not even possible. If you can provide some details such as what time controls you are using, if you are playing online and so on, I could suggest some settings that could help.

For online play, lag on the internet will require you to use a higher "Overhead ms" value. At least 100, but if your lag is high, use a larger number. The "Aggressiveness" parameter lets you also control how time is used up throughout the game. Look at the README.htm file for more information.

Time usage 9 works very well Mark, especially for 120/60 (I have played hundreds of games to back my point up; even at blitz Komodo seemingly will obviously use a lot more time but still plays as it should). What are you guys using for the TCEC tournaments? I think that you increased it a little bit right? I think that time usage always comes down to the belief of the author or user. Zappa and Houdini have horrendous time management, and we all know this. But SF and Komodo eat time for breakfast, lunch, and dinner. I would rather an engine lose on time than not use enough time.

[Modern Times]

No issues here either, it seems fine to me.

[carldaman]

Sudden death matches which involve no increments are ludicrous in my opinion, so steer clear of those.

Yes, I couldn't agree more. They're like a race to the bottom.

Regards,
CL

[lkaufman]

I want to thank you for your post, although it failed to make clear that you were referring to sudden death games (zero increment). Your post together with complaints from some users about time forfeits in sudden death games caused us to take another look at our time management for sudden death games, and indeed it was way too aggressive. On some GUIs (notibly Little Blitzer) Komodo 9 would time-forfeit over 10% of the games at game/1', which I consider unacceptable. Although on other GUIs it is not this bad, it is clear that Komodo 9 plays too slowly in sudden death games, and this lead to our release today of version 9.01. The reason this happened is that we always test with at least some small increment, and our time management works fine then. Version 9.01 plays sudden death games at about the same speed as Stockfish, although using a very different algorithm, so time usage should no longer be a significant factor in any Komodo 9.01 vs. Stockfish matches sudden death matches. As for the elo gain from version 9 to 9.01 in sudden death games, test results at different levels and different GUIs ranged from plus 2 to plus 18, with an average of about plus 8 elo. Most of the gain is due to (nearly) eliminating time forfeits, although results are slightly plus even without counting the forfeits.

[hgm]

How is that possible? If the games it forfeited would on the average have given it half a point, forfeiting 10% would have cost it 35 Elo. If it performs even better than the games it did not forfeit, the gain should even be more than 35 Elo. Was it predominantly forfeiting games it was losing?