Modern Times wrote:It would be good for Robert to confirm.
Robert did confirm in the on-line manual, and I quote: "You should only use one type of EGTB - if you specify several in the UCI options, the last option will prevail. Houdini cannot use more than one type of EGTB at the time."
Perfect.
Well, depending on the GUI you use, you sometimes cannot avoid more than one GUI path being sent to Houdini, without also affecting other engines. For the version of ChessGUI that I use, if I prevent Houdini from using Gaviota and Nalimiov, then that action prevents the opponent engine from using them as well - and the opponent engine of course doesn't have Syzgy. There really should be a UCI parameter in Houdini where you explicitly choose which ones you want to use and the GUI cannot affect that.
Question: Should I declare only one type of bases or all three (Gaviota, Nalimov, Scorpio) in the UCI-options?
You should only use one type of EGTB - if you specify several in the UCI options, the last option will prevail.
Houdini cannot use more than one type of EGTB at the time.
Modern Times wrote:It would be good for Robert to confirm.
Are you and Graham Banks brothers ?
Yes.
No way....
All these years I didn't notice that....
Come on, Doc
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
Id 241018 wrote:I have problems with houdini pro 4 in Aquarium as IDEA...freezes and slows down everything. Else anyone?
In ChessBase or Fritz, the cure is to set the engine to Below Normal priority, so that the engine does not hog all the resources and prevent even such basic things as allowing CPU power for the mouse.
"Tactics are the bricks and sticks that make up a game, but positional play is the architectural blueprint."
there is something broken in Aquarium, the GUI did not recognize all data send from the engine. As a quick fix Robert made an update, you should get it per email.
hgm wrote:A contempt different from 0 performing better is usually an artifact of biased testing, using many more opponents on one side of the engine than on the other. One should always pit the engine against a symmetric distribution of opponents, or the tests would become unreliable, and very dependent on the Elo model used.
Note that it is always possible to create stronger opponents by giving the opponents more time. Best is to test against set of opponents that score from 25% to 75% against the engine under test, equally distributed over that range.
Actually that is an excellent idea, although it sounds impractical for the top engines. The solution is simple: when testing Houdini, Stockfish, or Komodo, test them on 1 core against weaker opponents on 2 or 4 cores depending on the elo gap. It's not a perfect solution, as it is rather inefficient to test this way, but it should solve the problem mentioned here pretty well. In this way all one core engines can face a range of opposition centered at least not too far below their own level.
Dr.Wael Deeb wrote:The price of the new Houdini is reasonable considering the improvements advertised on it's web page....
Dr.D
P.S.I know that I'll be executed by Kim for this comment of mine
Hi Doc.
My guess is Houdini 4 is 70 elo. stronger than 2nd place engine.
Best,
Gerold.
I suppose you are talking about bullet chess and only official releases. Anyway H4 did end up beating h3 by 54 elo (+351 -197 =452) in my thousand game bullet match (30" +.3"), and is currently about 50 elo ahead of h3 at a slower 2' + 1.2" after about a thousand games (I'll actually run 4000 as I have 4 thousand game matches running on two monster computers). Allowing for some dropoff with longer time and when playing foreign opponents, I expect about +40 over h3 in the "new IPON" meaning the 5' + 3" CEGT list, which I think is overall the most meaningful list now considering both quantity of games and relevance of the time limit. This would put h4 about fifty above Komodo 6 or about 30 above our current version at this time limit. If so we're probably already equal with h4 at serious tournament levels but we have some work to do to catch h4 at blitz levels.
Gerolds guess was exactly right +70 above Komodo 6 on the "new Ipon"
Larry do you already have more testresults to make public ?
Dr.Wael Deeb wrote:The price of the new Houdini is reasonable considering the improvements advertised on it's web page....
Dr.D
P.S.I know that I'll be executed by Kim for this comment of mine
Hi Doc.
My guess is Houdini 4 is 70 elo. stronger than 2nd place engine.
Best,
Gerold.
I suppose you are talking about bullet chess and only official releases. Anyway H4 did end up beating h3 by 54 elo (+351 -197 =452) in my thousand game bullet match (30" +.3"), and is currently about 50 elo ahead of h3 at a slower 2' + 1.2" after about a thousand games (I'll actually run 4000 as I have 4 thousand game matches running on two monster computers). Allowing for some dropoff with longer time and when playing foreign opponents, I expect about +40 over h3 in the "new IPON" meaning the 5' + 3" CEGT list, which I think is overall the most meaningful list now considering both quantity of games and relevance of the time limit. This would put h4 about fifty above Komodo 6 or about 30 above our current version at this time limit. If so we're probably already equal with h4 at serious tournament levels but we have some work to do to catch h4 at blitz levels.
Gerolds guess was exactly right +70 above Komodo 6 on the "new Ipon"
Larry do you already have more testresults to make public ?
Yes. in a direct match Houdini 4 beat Houdini 3 by 58 elo at 2' + 1.2". When each played 4000 games against a recent Komodo, h4 outperformed h3 by 51 elo at 1' plus .5". This second figure is the more relevant for rating lists as they avoid self-play. Based on this, if there were no dropoff in the gain with greater depth, adding in the 16 elo advantage of h3 over k6 on the cegt list gives a difference of 67 elo over k6, very close to the 70 elo figure you cite. Since we have gained about 25 elo since k6 (at bullet levels anyway) we would need to gain 42 elo more to catch Houdini 4 at bullet chess. But we have probably already caught H4 at standard (40/2 hours) chess based on results posted here and on TCEC. This would also be consistent with my claim that we gain about 8 elo vs. Houdini every time the time control is doubled (obviously with some dropoff at high levels).
The only anomaly I have seen in the testing is that LS shows only +33 for h4 over h3, quite a bit below both my results and the CEGT result.
hgm wrote:A contempt different from 0 performing better is usually an artifact of biased testing, using many more opponents on one side of the engine than on the other. One should always pit the engine against a symmetric distribution of opponents, or the tests would become unreliable, and very dependent on the Elo model used.
Note that it is always possible to create stronger opponents by giving the opponents more time. Best is to test against set of opponents that score from 25% to 75% against the engine under test, equally distributed over that range.
Actually that is an excellent idea, although it sounds impractical for the top engines. The solution is simple: when testing Houdini, Stockfish, or Komodo, test them on 1 core against weaker opponents on 2 or 4 cores depending on the elo gap. It's not a perfect solution, as it is rather inefficient to test this way, but it should solve the problem mentioned here pretty well. In this way all one core engines can face a range of opposition centered at least not too far below their own level.
I would think giving the opponent more time is, for many reasons, far more practical than giving the opponent more cores.
