This game is probably won by Rybka, no doubt.
For the next match I will add information on the website that if for any reason an engine loses on time (ie crash or hangs or anything) it will lose and the next game begins.
That should clear it up.
Apologies about the decision I made regarding game 42 here, I wasn't aware that so many people thought the result should stand.
Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
Moderators: hgm, Rebel, chrisw
-
- Posts: 1833
- Joined: Thu Jun 22, 2006 12:07 am
-
- Posts: 8514
- Joined: Thu Mar 09, 2006 3:25 am
- Location: Jerusalem Israel
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
Martin Thoresen wrote:This game is probably won by Rybka, no doubt.
For the next match I will add information on the website that if for any reason an engine loses on time (ie crash or hangs or anything) it will lose and the next game begins.
That should clear it up.
Apologies about the decision I made regarding game 42 here, I wasn't aware that so many people thought the result should stand.
I also don't think it was all that terrible, after it was already conclusive who wins the whole match.
However, i think it is EVEN WORSE, if a game is decided because of a crash or similar.
There are possible alternatives, e.g. to reset from same positions and same clock times where it was.
With that, you could give a small time penalty to the side which crashed, to make sure that the side which did not crash should not lose thinking time and that the odds should be against the engine which crashed.
But, I don't think anyone in their right mind, really likes something to win or lose becaue of a crash. Even if all my heart was on one engine winning, and the other crashed, i would feel it wasn't the win i was hoping for, and would pretend to myself it wasn't played and i would even discard it from my feeling of what the real score is.
-
- Posts: 1167
- Joined: Thu Dec 25, 2008 9:07 pm
- Full name: Herbert L
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
Hi Ignacio,
thanks for the information, i've overlooked that.
thanks for the information, i've overlooked that.
-
- Posts: 1833
- Joined: Thu Jun 22, 2006 12:07 am
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
That was my initial reaction as well, I don't enjoy it if an engine gets a full point because of a crash of the opponent.S.Taylor wrote: There are possible alternatives, e.g. to reset from same positions and same clock times where it was.
With that, you could give a small time penalty to the side which crashed, to make sure that the side which did not crash should not lose thinking time and that the odds should be against the engine which crashed.
But, I don't think anyone in their right mind, really likes something to win or lose becaue of a crash. Even if all my heart was on one engine winning, and the other crashed, i would feel it wasn't the win i was hoping for, and would pretend to myself it wasn't played and i would even discard it from my feeling of what the real score is.
But the problem with your suggestion is that it's not possible to do, if an engine crashes and loses on time, the database records victory for the other engine and the next game starts. So if I stop and replay the game that was lost on time, the position starts from the beginning. This is just how the Chessbase GUI is created and there's nothing I can do about it.
Unfortunately.
-
- Posts: 3707
- Joined: Thu Mar 16, 2006 7:10 pm
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
Rybka 4 default time management with CCRL and CEGT setups is terrible. It will spend way too much time on a few moves and quickly end up in terrible time trouble. We saw that in beta testing but had to come up with a TC setup that was a good compromise for all in the little time we had. I remember getting results against SF 1.71 that would change more than 80 ELO by simply changing one TC parameter. It is also one of the reasons we requested that the TC parameters be left available to change. The default TC setups that R4 especially had trouble with was when you had say 40 moves to play in X amount of time, or in very fast lightning games, so I think you will find that R4 will underperform in those setups when compared to games where there is increment or in LTC games where poor TC management is not as big an issue. Hopefully this will be improved with the R4 update. I wouldn't be surprised if with the R4 update that R4 will score 40+ ELO higher than what it is scoring now against SF in CEGT even though chesswise it will be the same engine.Dann Corbit wrote:Considering the beatdown that Stockfish 1.8 is having, I suspect that something is wrong with your configuration. Have you verified (for instance) that Stockfish is using all the threads you want it to use?
The reason I suspect a setup snag is the following:
http://www.husvankempen.de/nunn/40_40%2 ... liste.html
-
- Posts: 8514
- Joined: Thu Mar 09, 2006 3:25 am
- Location: Jerusalem Israel
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
If so, i would vote for adjudication (if it is clear enough), or, replay (esp. if it is not clear).Martin Thoresen wrote:That was my initial reaction as well, I don't enjoy it if an engine gets a full point because of a crash of the opponent.S.Taylor wrote: There are possible alternatives, e.g. to reset from same positions and same clock times where it was.
With that, you could give a small time penalty to the side which crashed, to make sure that the side which did not crash should not lose thinking time and that the odds should be against the engine which crashed.
But, I don't think anyone in their right mind, really likes something to win or lose becaue of a crash. Even if all my heart was on one engine winning, and the other crashed, i would feel it wasn't the win i was hoping for, and would pretend to myself it wasn't played and i would even discard it from my feeling of what the real score is.
But the problem with your suggestion is that it's not possible to do, if an engine crashes and loses on time, the database records victory for the other engine and the next game starts. So if I stop and replay the game that was lost on time, the position starts from the beginning. This is just how the Chessbase GUI is created and there's nothing I can do about it.
Unfortunately.
But not to count irrelevant happenings.
-
- Posts: 1833
- Joined: Thu Jun 22, 2006 12:07 am
-
- Posts: 1833
- Joined: Thu Jun 22, 2006 12:07 am
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
At the half-way mark (16 of 32 games) the score is 9.0 for Houdini vs 7.0 for Stockfish.
-
- Posts: 10121
- Joined: Thu Mar 09, 2006 12:57 am
- Location: van buren,missouri
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
Update- 14.5 Houdini. 11.5 Stockfish.Martin Thoresen wrote:At the half-way mark (16 of 32 games) the score is 9.0 for Houdini vs 7.0 for Stockfish.
Houdini is having fish for dinner
Best,
Gerold.
-
- Posts: 543
- Joined: Mon Jul 05, 2010 10:27 pm
Re: Stockfish vs Houdini - Long TC Match - LIVE BROADCAST!
Game 27:
Looks like Houdini will won this one for a 15.5 - 11.5 score
Not big difference at all
I like Houdini because its free, but Im biased towards Stockfish because its
open source. They are true free and has the disadvantage of ideas visible
for the ones making non free programs. and this is a more than 50 ELO
points handicap for sure
Go Stockfish
Looks like Houdini will won this one for a 15.5 - 11.5 score
Not big difference at all
I like Houdini because its free, but Im biased towards Stockfish because its
open source. They are true free and has the disadvantage of ideas visible
for the ones making non free programs. and this is a more than 50 ELO
points handicap for sure
Go Stockfish