New engine test for the IPON list is running

[IWB]

Hello

It's been more than 6 months that a 'loosly called' clone appeared. The accusation - right or wrong -
is, that it is a decompiled Rybka 3.
Within these 6 months I collected every littly hint which might lead to some criminal act. As the
programmer of Rybka did not give any evidence (which is understandable as it is difficult) and I am
not an expert on programming I asked a dozent programmers about their thoughts on that source code.
I got 9 answers with a lot of indications - but no proof. Still I had and have a bad feeling about the
engine. On the other hand it is for sure more than a 'simple' decompile. It seems to be understood and
enhanced in many ways. I personaly think that this has a bad taste, if it is illegal might vary from
county to country.
The 'original clone' intensionaly does not have a real, known person as an author which make it difficult
if not impossible for someone to claim it's intellectual property but there are some new enigenes which
openly admit their use of the doubtfull sources. At the same time there are more and more enignes emerging
with remarkable increases in playing strength in a very short time and as a very strong source code is
available the assumtion that some of these enignes have borrowed something from that source code is likely.
All rating lists are already testing and supporting these engines as they do not have any other possibility
besides stop testing enignes at all.
So we have capable programmers who openly admit to use the source and we have a few who dont, some use more
of the code some less.
To end this dillema I decided to test one engine of a person who, right from the beginning, never tried to
hide the origin of his engine, who seems to be a professional programmer without to much of an ego. Additionaly
it seems that no one tried to stop these engine despite the fact that everything about the author is known. I
will remove the engine from my list the day I hear something different.
I will only allow one of those enignes, which individually might be changed in the future, in my list to
avoid having to play 10 nearly identical motors.

Current IPON Test: http://www.inwoba.de

Please be aware that the actual elo calculation on my web site is provisional. It assumes a fixed elo rate for the opponents, which of course is faulty. The final elo number might be different!

Bye
Ingo

[kingliveson]

It is overdue. Thank you.

[mcostalba]

Hi Ingo,

this is a very important choice from your side, and I second that.

I also think you have picked up the correct one, not because it is stronger, but because of the good "attitude" of the author. Actually he admits that he has used Ippo, SF and Crafty as references, well, my grunt feeling (but no proof) is that he actually "started from" Ippo sources and then added ideas from other engines, but I consider this not of paramount importance.

Many people started from other engine's sources, even the current official number 1 is rumored to have started (and gone commercial !) from another engine's sources, and that case is also worst because that engine was GPL and so should have been the result...but has been not.

It is a pity I cannot give a look at the resulting sources. Having started from public domain sources it is a "natural" choice to post also the resulting ones (not legally enforced though because of public domain nature of Ippo). On the other hand if what I would find is Ippo + some SF / Crafty ideas thrown in well I am not so interested, nothing new. If instead the author put together some original ideas then of course things get more interesting.

Anyhow you have opened a door that others will follow....

Best
Marco

[Swaminathan]

Thanks, I'll follow your ratings list. It's entertaining and interesting.

[IWB]

Houdini 1.02 test finished.

Be aware that the enigne is controversial of beeing a "clone/derivate"!
Please inform yourself in the computerchess fora to judge about it's status.

http://www.inwoba.de

Regards
Ingo

[Eduard Nemeth]

Houdini 1.02 won against the two leaders (Rybka 4 and Stockfish 1.7.1) with more than 50%!

Code: Select all

Houdini 1.02 - Deep Rybka 4        (2948)   51.5 - 48.5  51.50%  Perf=2958 
Houdini 1.02 - Stockfish 1.7.1 JA  (2882)   59.5 - 40.5  59.50%  Perf=2948

Fantasic result!

[Wolfgang]

Houdini 1.02 won against the two leaders (Rybka 4 and Stockfish 1.7.1) with more than 50%!

Code: Select all

Houdini 1.02 - Deep Rybka 4        (2948)   51.5 - 48.5  51.50%  Perf=2958 
Houdini 1.02 - Stockfish 1.7.1 JA  (2882)   59.5 - 40.5  59.50%  Perf=2948

Fantasic result!

Hi Eduard,

really fantastic that Houdini wins with "more than 50%"! Can you tell me, how someone can win with less than 50% or even with exactly 50%??

I just make some tests with Houdini 1.02 x64 1CPU too. Conditions from CEGT but the test is inofficial / private because CEGT decided not to test "these" engines at the moment!! Time will tell, if (and how long) we can hold this point of view....

First match vs. Naum 4.2 x64 1CPU ended: 141,0 : 59,0 = 70,5% = +151 ELO (Rybka 4 socred 71,7% vs. Naum)

Second match is running vs. Stockfish 1.7.1 x64 1CPU.

All matches on
* Intel Dual E5200 (2*2,5 GHZ)
* 40 moves / 3 minutes repeated
* Shredder Classic GUI
* 200 games per match
* testsuite from Heinz van Kempen with 100 positions

Best wishes
Wolfgang