1
2
![[mx]](/mwf/v12/flags/mx.png "Mexico")
> 1) The flattened H2 scores make them almost useless for IDeA used interactively;
Thanks for confirmation that H2 shows the same symptoms of Houdini 1.5. I claim that this is true even if one doesn't use IDEA but uses Infinite Analysis interactively.
>R4 has an unfortunate tendency to maintain scores into the endgame that are much too high
Also true. But any engine that you can name has some kind of major problem somewhere, that's why I claim analyzing with several engines is best.
> 1) The flattened H2 scores make them almost useless for IDeA used interactively;
Indeed, it's not easy to combine in IDeA results from engines that have different score calibration (Stockfish and Houdini being at the extremes of the scale).
That's why the Houdini 2 Aquarium package contains a special single-core Houdini 2 IDeA version with slightly inflated scores, this should remove the potential problem with Houdini 2 in the IDeA.
> the Houdini 2 Aquarium package contains a special single-core Houdini 2 IDeA version with slightly inflated scores
Inflating scores doesn't solve anything when the problem is that Houdini gives the same score to many moves.
You change a flat line
________ for a flat line --------, which is equally ineffective.
> Inflating scores doesn't solve anything when the problem is that Houdini gives the same score to many moves.
This whole premise of yours is incorrect, and in fact doesn't make a lot of sense. No engine can be the most accurate or successful in finding the "best move" without being also the most accurate or successful in assessing the alternative moves.
I don't want to turn this into an endless debate - our perspectives are very different so it's OK if we disagree. I accept your observation that Houdini doesn't seem to fit your needs, hopefully future versions will change that :).
> I don't want to turn this into an endless debate - our perspectives are very different
I'll play the experience card on this. How much experience do you have analyzing positions for correspondence games with engines? And how much experience do you have doing that with many engines?
Because, you may know a lot when it comes to programming, and when it comes to improving the engine to maximize its elo and for making it beat other top engines and even also beat weaker engines, but this "doesn't make a lot of sense" from you would make sense depending on the experience you have with chess engines.
It is claimed that Rybka 4.1 is better than Houdini for analysis, how much have you used Rybka 4.1 for analysis to claim that Houdini is better? Because otherwise you are just speculating ("good game results = best analysis") while people claiming Rybka is better after having used both are talking about their experience.
Like I said before your previous post.
I don't want to turn this into an endless debate - our perspectives are very different so it's OK if we disagree. I accept your observation that Houdini doesn't seem to fit your needs, hopefully future versions will change that :).
I don't want to turn this into an endless debate - our perspectives are very different so it's OK if we disagree. I accept your observation that Houdini doesn't seem to fit your needs, hopefully future versions will change that :).
![[in]](/mwf/v12/flags/in.png "India")
IMHO the learning feature of Houdini is not as effective as Critter. You if can change that then It will be an hell of an analysis tool.
![[us]](/mwf/v12/flags/us.png "United States")
Any chance you can give us some details in this thread about Houdini 3?
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=24448
I think a lot of us are interested in what you have to say about
"it will do quite well at long TC - some search algorithm improvements will have an increasing effect with higher search depths."
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=24448
I think a lot of us are interested in what you have to say about
"it will do quite well at long TC - some search algorithm improvements will have an increasing effect with higher search depths."
![[jo]](/mwf/v12/flags/jo.png "Jordan")
What do ecpect him to say else !?
In the end,he must do some advertising regards,
Dr.D
In the end,he must do some advertising regards,
Dr.D
Your right, I know there is some sells tactics in there. I thought maybe would give him a chance to add something to it. I don't think he is full of hot air all the time.
![[ch]](/mwf/v12/flags/ch.png "Switzerland")
I used Houdini and Rybka a lot for analysis and CC.
Houdini is just stronger and more efficient.
I use it as my main engine and Rybka and Komodo as 2nd opinion
Houdini is just stronger and more efficient.
I use it as my main engine and Rybka and Komodo as 2nd opinion
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
Quite easy, loosing games with H2 that Rybka did not lose this was over many months of CC including H1.5.
IMO H2 is the king at Blitz but CC time controls no way.
IMO H2 is the king at Blitz but CC time controls no way.
see my reply to your earlier post.
>Quite easy, loosing games with H2 that Rybka did not lose this was over many months of CC including H1.5.
Otherwise is also true.
>IMO H2 is the king at Blitz but CC time controls no way.
Are u serious ?
. Whats the time You tend to Run Rybka ? 4 hours ? or more ? or in IDeA ?
Yes i am serious. Depends on the position but on average 6-8 hours with human interaction. Never used IDeA.
that was my opinion until recently,it depends upon the position,i found in my WBCCC game against Schachmatt just finished in a draw that Houdini was more useful than Rybka.
I havent read all the posts after this reply,but i talk as an experienced Corr player.
Bet Robert loves this post as it is a bit of a climb down by me.
I havent read all the posts after this reply,but i talk as an experienced Corr player.
Bet Robert loves this post as it is a bit of a climb down by me.
Houdini was more useful than Rybka in one out of how many games where you're used it?
(Question meant to show that Houdini over-performing Rybka is more the exception than the rule)
(Question meant to show that Houdini over-performing Rybka is more the exception than the rule)
i take your point but in game above it was true.
Uly,
I see you are challenging a lot the ones who claims that Houdini is currently better than Rybka, but on the other hand bring no data to prove the contrary which you claim.
Rgds
I see you are challenging a lot the ones who claims that Houdini is currently better than Rybka, but on the other hand bring no data to prove the contrary which you claim.
Rgds
I could give some practical examples of this general scenario:
Position is quiet, there's many plans that the player can execute on the board. After interactive analysis, Houdini gives all plans a score very similar, leaving the user without a firm resolution of what to play. A different engine has a favorite plan that is useful to the user.
What to play in such positions is very important, as it's the only way to reach the ones where the opponent will play inaccuracies.
Just let me be up to date with my corr games and I'll do this.
Also, you could post a position in where Houdini is the only engine that suggests a plan that is best, that no other engine would have seen after interaction, as evidence that "Houdini is just stronger and more efficient", because the only evidence is game results and the correlation between game results and analysis is unclear (Rybka 3 has been surpassed by engines in the elo department but I still find it as useful to analyze positions with as Rybka 4.1).
Position is quiet, there's many plans that the player can execute on the board. After interactive analysis, Houdini gives all plans a score very similar, leaving the user without a firm resolution of what to play. A different engine has a favorite plan that is useful to the user.
What to play in such positions is very important, as it's the only way to reach the ones where the opponent will play inaccuracies.
Just let me be up to date with my corr games and I'll do this.
Also, you could post a position in where Houdini is the only engine that suggests a plan that is best, that no other engine would have seen after interaction, as evidence that "Houdini is just stronger and more efficient", because the only evidence is game results and the correlation between game results and analysis is unclear (Rybka 3 has been surpassed by engines in the elo department but I still find it as useful to analyze positions with as Rybka 4.1).
Why dont you just go your own through these different plans which seems similar in evaluation and use your own brain and feelings to play what you as a human prefer or think is best since the best engine is giving no clear favour ?
I dont believe either that folling blindlessly the engines is the best and likely doesn't procure more pleasure than a more active participation of your own brain & tastes.
Looking forward to see your positions & data.
If you look today available data (rating lists) Houdini is the best performer.
The only situation I agree of using a different engine than the best performer makes sense, is that some CC players are some much used to analyse with a specific engine, that they know/feel when the engine gets it wrong or when one cannot relay 100% on the engine output.
A friend of me is about 2550 ELO in CC and use Fritz 11. He is reluctant (for how long?) to change since he knows really well when the engine goes wrong (most of the time?)
I dont believe either that folling blindlessly the engines is the best and likely doesn't procure more pleasure than a more active participation of your own brain & tastes.
Looking forward to see your positions & data.
If you look today available data (rating lists) Houdini is the best performer.
The only situation I agree of using a different engine than the best performer makes sense, is that some CC players are some much used to analyse with a specific engine, that they know/feel when the engine gets it wrong or when one cannot relay 100% on the engine output.
A friend of me is about 2550 ELO in CC and use Fritz 11. He is reluctant (for how long?) to change since he knows really well when the engine goes wrong (most of the time?
)
> Why dont you just go your own through these different plans which seems similar in evaluation and use your own brain and feelings to play what you as a human prefer
I do that and in the end I'm the guy deciding, but what if all the plans look fine and playable? There's got to be a reason for going for a plan and it should be stronger than just "I like that variation because the king is more active than the other variations".
One doesn't even need to explore the plans for that, one can simply look at the PV and visualize how the lines develop. This leads to much a weaker plan than engine crosschecking, though, and for this, it's important that an engine has a preference for some line and shows the user why it's the best course of action.
Exploring the variations with Houdini is moot if it doesn't show the user the line that would convince it to go for it. If Rybka Dynamic clearly prefers the move and shows a nice looking variation that the human feels it's better, while Houdini shows dull lines for all moves, I'd rather use Dynamic.
Here's an actual example or a recent game in the WBCCC:
1.Nf3 c5 2.c4 Nf6 3.Nc3 g6 4.g3 Bg7 5.Bg2 O-O 6.O-O Nc6 7.d4 cxd4 8.Nxd4 Qb6 9.Nc2 d6 10.b3 Qc5 11.Bb2 Qh5 12.e4 Bg4 13.f3 Bd7 14.Qd2 Rac8 15.Ne3 Qa5 16.Rad1 Rfe8 17.Ncd5 Nxd5 18.exd5 Qxd2 19.Rxd2 Nb4 20.Bxg7 Kxg7 21.Re1 a5 22.Bf1 h5 23.Kf2 Kf6 24.Nd1 Na6 25.Nc3 Nc5 26.Re3 Ra8 27.Be2
On this position I examined many plans of continuation by black, and at some point, Rybka 3 Dynamic was enthusiastic about the plan with Ra6-Rb6, and showed a couple of variations where Black would get counter-play if White just shuffled its pieces. Other engines disagreed and ended up refuting the move, so that no engine would like Ra6 and even Dynamic would prefer a different move. But none of the other variations that I saw convinced me to go for them, so I played Ra6.
In the actual game it turned out my opponent went into positions where playing Rb6 would have been bad, and I never got to play it, but in the process he allowed the e5 push that led to full game equality (Rybka 4.1 still giving 0.30 something to the actual position but I can refute all its ideas on demand). Ra6 was the best move as there was the only one with a reason to play it.
I never used Houdini on this game, but I wonder, would it be enthusiastic about any move at all? Or would it be scared by white edge and just rearrange the deck chairs in the titanic?
I think it depends on how u use it. Houdini is a class in finding awkward and solid defences. Thats why its the top engine.
Like if you have a fort no one can break easily, U have won the war already without fighting.
Well in that game Stockfish was better than the two.
Like if you have a fort no one can break easily, U have won the war already without fighting.
Well in that game Stockfish was better than the two.
Maybe i have not give H2 a good enough try?
Anyway Paul it's only my opinion which does not count for anything.
i generally prefer to use Rybka but on occasion Houdini is better because of fast exe.
![[ar]](/mwf/v12/flags/ar.png "Argentina")
Hi.
Look my config:
http://computerschess.blogspot.com/
If you want buy it i can sale.
I no need now it monster.
Look my config:
http://computerschess.blogspot.com/
If you want buy it i can sale.
I no need now it monster.
Have you since got a new computer to add to your collection?
No, i no need more computers. I am using the 980; 950; 920; Phenom and i5. It is ok for corr chess. I no play on Playchess, no need 12 cores, and need money.
tienes 4 ordenadores aparte del de 12 cores para jugar al ajedrez,y todos ellos ultimos modelos,y ademas el 12 cores?
edit:you have 4 computers appart from the 12 cores to play chess,and all of it latest models,and also the 12 coes computer?
edit:you have 4 computers appart from the 12 cores to play chess,and all of it latest models,and also the 12 coes computer?
Lejos están de ser últimos modelos.
No los tengo sólo para jugar al ajedrez.
No los tengo sólo para jugar al ajedrez.
hombre,poco se de ordenadores,pero 97'y pico que suenan todos,deben ser todos de 4 procesadores,y el i5,es de 4 procesadores si no me equivoco tambien,aparte del monstruo de 12 procesadores...ya querria tener yo solo el i5,fijate lo que te digo
viste a las hijas del ex presidente Español (Zapatero) cuando las llevo a la recepcion oficial con Obama?
juro por dios que noo es un montaje,son asi de ******,bueno,digamos que son asi...
el video sale al principio la foto oficila de la recepcion,y luego ya empiezan los montajes,pero se notan enseguida porque estan caricaturzados...pero la foto es de una recepcion oficial,y el ****** de nuestro ex presidente las llevo a una recepcion oficial asi vestidas...
dime tu si no es para atar por el cuello al Zapatero a un coche en marcha y arrancar el coche...ni vergüenza tiene llevar a esas ***** a una recepcion oficial asi...,ira el video y dime que opinas
http://www.youtube.com/watch?feature=endscreen&NR=1&v=UVkBdv-Qnhw
juro por dios que noo es un montaje,son asi de ******,bueno,digamos que son asi...
el video sale al principio la foto oficila de la recepcion,y luego ya empiezan los montajes,pero se notan enseguida porque estan caricaturzados...pero la foto es de una recepcion oficial,y el ****** de nuestro ex presidente las llevo a una recepcion oficial asi vestidas...
dime tu si no es para atar por el cuello al Zapatero a un coche en marcha y arrancar el coche...ni vergüenza tiene llevar a esas ***** a una recepcion oficial asi...,ira el video y dime que opinas
http://www.youtube.com/watch?feature=endscreen&NR=1&v=UVkBdv-Qnhw
i understand Ruben,the more power you get the more you realise how complex the game is,i have posted before that in certain positions half a dozen computers would be useful.
well,so maybe is a blessing be a poor guy and have only an old computer so im not able to see how complex is the game and i can focus in some of other things of the real life
i cant understand people with sooooo much computers just to play chess:that is an obsession,and that must not be good
i cant understand people with sooooo much computers just to play chess:that is an obsession,and that must not be good
i did not mean that.i meant what i said.Chess is so complex that many computers would be of little use,this is where human judgement takes over.
i was joking Paul
how much you are selling it?
I listen an offer.
1
2
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill