An Old Engine

[supersharp77]

with a new name , that now , sadly, falls under the Stockfish GPL.
More discussion may be found here:
http://talkchess.com/forum3/viewtopic.php?f=10&t=73433

Houdidit Release 6.03
https://github.com/MichaelB7/Houdini/releases/tag/6.03

Bench:

houdidit-x64-amd bench 2048 64 20
Houdidit 6.03 Pro x64-popc
Stockfish Developers & Robert Houdart

info string NUMA configuration with 2 nodes, offset 0
info string NUMA node 0 in group 0 processor mask ffffffffffffffff
info string 32 cores with 64 logical processors detected
info string 1 thread used
info string 128 MB Large Page Hash
info string 2048 MB Large Page Hash
info string 64 threads used
39 bench positions


Total time (ms) : 113316
Nodes searched : 18627286115
Nodes/second : 164383000

I was always taught that, "two wrongs don't make a right."

I am now pleased to make the source available to everyone, as it is my opinion that the allegations are true and the evidence provided is overwhelming.

We are all allowed our opinions and are free to express them and we should encourage that. But I think I mentioned in the Engine Origins post, I don't agree that people should go further and act as judge, jury and executioner in cases like this. Unfortunately in the chess community that seems to be normal.

I bought every version of Houdini ever sold. I am feeling no sting from it.
The exchange of information wrought by an engine being open source is intended, and indeed almost every strong engine borrows concepts from other strong engines.
Of course, we should do this without breaking the law (goes without saying).
I never have buyer's remorse.
For instance, I spend a lot of money on hardware for chess and other things.
If I sat and waited, I could have spend exponentially less money and got the same power.
Or I could have spent the same money and got exponentially more power.
But in the meantime, I had a wonderful machine that did everything I asked of it and it made me happy, not sad.

Similarly with Houdini. I bought Houdini because it is extraordinarily good a solving tough chess problems that other computer chess programs may have difficulty with. During the time that I had it, I used it to analyze literally millions of chess positions.
Now I see that Mr. Houdart may have made illegitimate use of code that did not belong to him.
But that changes nothing about the benefit to me from using his program in the past.
Better still, because the code is based on Stockfish, I will be able to splice in some of my code to write the analysis directly to disk in a format that is easy to load into my database systems. Hey, double bonus.

Getting your bun in a knot never accomplishes anything. Sure, we all do it from time to time, but what good came of it?

I agree completely that the SF team is the injured party.

I don't really want to get into this discussion, but I have a feeling if you would ask someone like Marco Costalba, that he would expect it to happen with Stockfish code because for instance the GPL does not give much protection. Fruit being a case in point. And I think he does not care super much, even if he was still very active, nor would the other developers, well most of them I think. But that is just a guess of mine. Stockfish gets its advantage from the continuing development and testing and lone developers have to build in all sorts of copy protection, have to deal with buying customers, have limited testing hardware, very limited development time as single developers, doing everything on their own etc. Is it not much more Mark Lefler and Larry Kaufman, and maybe a few others who are still making commercial chessprograms, with their own ideas, who are the injured party here? That was my first reaction. What others do wth the Stockfish code, well it will be hard to do better than Stockfish even if they copy it a 100% I would not care about infringement of the GPL, of course it is wrong but it is very hard to hide and using the ideas, rewriting algorithms is not protected anyway. This is more a sportsmanship question for what I care.

Sorry My Friends....Tried My Best But By The Time I realized what was going on.."The Horse Had Already Left The Barn and Was Rampaging In The Town Square"....I missed entirely the subtle post "An Old engine" and rarely peruse the "Engine Origins Section" choosing to bypass the "Rybka vs Fruit Or Robbo vs Houdini Or Houdini vs Fruit Arguments"
My understanding of the history is...Fruit led to Glaurung...Glaurung led to Stockfish and Almost all follow what Stockfish is doing (Including Alpha Zero & LC0) Also Fruit led to Robbolito. Robbolito was used by Houdart to inspire Houdini
So Perhaps Fruit should be getting way more credit than has been given out so far??...Meanwhile Chessbase usable versions of "Houdidit" flood the internet...expect this unfortunate controversy to persist for a good while until Houdart shows up to give his opinion on the matter which I expect he will....

[chysiddh14]

i m not a chess expert nor a programer but i have seen this houdidit chess engine source code month ago in github.
MikeB has renamed stockfish fork to play like houdini ?
thats why he rename houdidit instead of houdini .

[chysiddh14]

1 more thing to add.
i have seen this houdidit chess engine play in fics or lichess bots i m not sure . but i have seen this chess engine months ago trust me ,i m telling truth.
in description it has written it is stockfish fork to play like houdini.

[MikeB]

i m not a chess expert nor a programer but i have seen this houdidit chess engine source code month ago in github.
MikeB has renamed stockfish fork to play like houdini ?
thats why he rename houdidit instead of houdini .

To be clear:
No, not true at all. I'm not aware of any SF fork that was renamed or Houdini or Houdidit . Certainly not by me. I had Git standalone repos that were named Houdini and Houdidit. These two similar but different sources were downloaded from other repos and I have no first hand knowledge of the source of such repos other than what I have read on various forums which may or not be true.

Since I have no first hand knowledge of how these sources were obtained, I elected to remove my repos named Houdini and Houdidit on Github earlier today and be done with it.

Bye.

[Ovyron]

Since I have no first hand knowledge of how these sources were obtained, I elected to remove my repos named Houdini and Houdidit on Github earlier today and be done with it.

I want to applaud MikeB for his original release and wish that he reconsiders putting the sources back up.

I've been wishing really hard for a release of Houdini 7 or a continuation of the project, if you search for my posts on here or Uly's posts on Rybka Forum you'll see I had nothing to say but nice things about the engine's playing style. I was heartbroken by the news of Robert Houdart quitting computer chess because that meant the dream was over, but there was a glimmer of hope when it was reported that someone had gotten the engine's source code and that it could potentially be released.

We don't need Houdart for a Houdini 7, or for a Houdidit 2, or the way this ends being called. It's a source code that exists, and if we were humans from 200 years from the future that found the code and wanted to work on it, there wouldn't be a reason from doing so. I really hope someone picks it up and continues working on it, hopefully keeping it open source, but I had already planned to get a commercial Houdini 7, and how the code came to be makes no difference.

The community as a whole can benefit from all the original ideas in the code, like for instance, the way Contempt works, which is unique and very unlike Komodo or Stockfish's contemtps, so I'd love to see this style implemented in other engines.

I find it's a shame how MikeB stepped back from the releasing just because someone posted something (chysiddh14 must be very powerful, with the ability to make others delete githubs...) and hope that someone in the future rectifies this injustice (it's not fair that only people that were on Talkchess when the release happened got the code, while someone like me wasn't, so though luck because after one blinks, the source is gone...)

[Terje]

...
I find it's a shame how MikeB stepped back from the releasing just because someone posted something (chysiddh14 must be very powerful, with the ability to make others delete githubs...) and hope that someone in the future rectifies this injustice (it's not fair that only people that were on Talkchess when the release happened got the code, while someone like me wasn't, so though luck because after one blinks, the source is gone...)

The source is available.

https://groups.google.com/forum/#!topic ... ygaIdBvJm0

[Ovyron]

The source is available.

Thanks. For the convenience of the general public that wants the code without having to jump through hoops, here's a compilation of some known sources:

https://www.chessdb.cn/downloads/shame.png (yes, that's the source, you need to download the image and change the extension to a z7 file, then unzip it) Credit: noobpwnftw (all my applause goes directed to him as it seems all other sources come from here)

https://github.com/alureon/houdini (Credit: Andrea Manzo)

https://www.dropbox.com/s/h9w3pankltjqv ... d.zip?dl=0 (Credit: Alexey Eromenko)

You may have problems because of a missing file, try this:

Just comment out all if defines with the word license , I also made a blank file called 'licentie.cpp"

###feq ($(VERSION),Std)
# CXXFLAGS += -DPRO_LICENSE
#endif

BONUS:

The Stockfish from where Houdini 5 was based on:
https://gitlab.com/cucumbers/stockfish-frozen

I really hope someone does something cool with this, if this is the end of the line it's as if nobody ever made the discovery, this could be the Ippolit of our generation, something needs to happen for a happy ending.

[Vinvin]

I found a buggy behavior on Hashtbales size :
Often when I set a hash size on a low value (2048 MB or 1024 MB), the memory used reported by the Taskmanager is very low (10 MB to 15 MB).
Not always but very often.

Could someone confirm this, please ?

[Ras]

Often when I set a hash size on a low value (2048 MB or 1024 MB), the memory used reported by the Taskmanager is very low (10 MB to 15 MB).

Does this persist when running e.g. analysis for a few seconds after changing the hash size? Or does the usage go up?

[Vinvin]

Often when I set a hash size on a low value (2048 MB or 1024 MB), the memory used reported by the Taskmanager is very low (10 MB to 15 MB).

Does this persist when running e.g. analysis for a few seconds after changing the hash size? Or does the usage go up?

It still at 10-15 MB. Even after 2 minutes of analyze. Despite the engine display "2048 MB Hash".

Code: Select all

Houdidit-modern:
Houdidit-modern
2048 MB Hash
Syzygy 6 men EGTB available - 510 tablebases found
Nalimov EGTB not available