Questions on SMP search

[bob]

Hi Roger,

It cannot be that here, in the Technical Sub-forum, that we are going to conduct arguments along the line of who are you, how long have you been involved in computer-chess and is Houdini an original program.

It is also not going to be tenable that wherever you post that you can be asked is Houdini original etc. Then every thread could potentially spiral out of control.

If even Bob feels perfectly happy to produce false quotes and personal attacks, I don't think the anti-Houdini trolls will feel any restriction in the future. It will strongly limit my willingness to interact with this forum.

"Limit your willingness to interact"???

If there is evidence produce it and let it speak for itself.

I have little evidene unless I publish the source code, which I don't intend to do now.
And against people that are happy to produce false quotes to discredit me, even that might not be good enough.

Thanks for your much appreciated moderation effort,
Robert

Have you ever claimed that Houdini was original? Have you ever claimed it was derived from another program or programs? Both of those can't be false statements, now can they? Did you respond to C. Conkies posts showing _identical_ output produced by Houdini and Robo* for a series of positions? Exactly how likely is that when we are not talking about mate in N or win in N type positions?

I am personally convinced, 100%, that you are simply a fraud. Yes, you managed to improve Robo* in a fairly significant way. That doesn't change a thing, however....

[bob]

It's funny how some things stand out, like this:

rec_stelling_var *VAR_0 = VAR;

compare to:

tipo_dinamico *POSIZIONE0 = DINAMICO;

at the same place in the RobboLito sources.

I don't think it is funny at all...

[Houdini]

Oh my Gosh ! You have chosen variable names in your own mother language !!! That's very funny, it remembers me of Roberto Pescatore's italianized Ippo sources but there I think a good chunk of irony was involved...I am not sure the same is here.

LOL, I should have called myself Bob Visser then?
My chess developments have always been in Dutch (which is indeed my mother tongue), it's a very nice chess language .

Robert

[Karlo Bala]

Let me show you an actual code snippet of a function definition:

Code: Select all

template<$$def_THREAD_ID$$ Kleur IK, bool MULTICORE> 
int Knoop_All($$def_STELLING$$ int BETA, int DIEPTE)
{
	const Kleur JIJ = (IK == WIT ? ZWART : WIT);
	rec_stelling_var *VAR_0 = VAR;
...

Oh my Gosh ! You have chosen variable names in your own mother language !!! That's very funny, it remembers me of Roberto Pescatore's italianized Ippo sources but there I think a good chunk of irony was involved...I am not sure the same is here.

All my code is written in Serbian. What is funny about that?

[Houdini]

Concerning the following structures of your chess engine:

- Main Hash Table
- Pawn Hash Table
- Killers array
- History array

How you deal with these structures, when implementing a SMP search? To be more specific:
- For what structures, you keep a separate instance for each thread and what structures share a single instance with all threads;
- How you handle the concurrent access - if applicable - for each of these structures;
- How to maintain the killers/history arrays.

Thanks,

Ben-Hur Carlos,

I'm really sorry that my presence has generated a complete hijacking of your thread.
I'll make no further comments in this thread, as everything I say seems to work like a red flag on a bull.
Hopefully you'll be able to resume the thread for its intended purpose.

Have a nice day,
Robert

[mhull]

3) I've been involved in computer chess since around 1985. That makes about 25 years, not your "over 30 years".

Involved? Many people say this because they owned a chess computer or a program and played with it. It has the sound of that guy who once boasted on this board that he had worked on the SETI project, which it turns out amounted to running the SETI@Home screen saver on his home computer for a while.

I don't suppose you'd care to elaborate, Bob Hyatt style, on the nature of your computer chess involvement in 1985.

My involvement has been on a personal level, I've been studying and writing chess related code since about 1985 (starting with Richard Lang's Psion chess on a Sinclair QL).

You seem to suggest you hacked the Psion code IPPO style, to perhaps copy and modify to make your own thing, Houdini style.

My own efforts were mostly in assembler, turbo pascal, Delphi. I have in the past sent two of my earlier engines + GUI (written in assembler + Delphi) to a couple of forum members, send me an e-mail if you want your own copy.

You could send them to the Chess Programming Wiki who seem unaware of these formative achievements. You could send a copy to Mr. Watson for instance.

Note that I've already stated all this at least 2 or 3 times before on this forum, if you'd care to use the forum search function you'll find more information.

I've seen them and they are vague like now.

And again, what all this has to do in a "Question on SMP Search" thread is a mystery...

You should answer this yourself since I'm only responding to a claim YOU made.

[Laskos]

I am personally convinced, 100%, that you are simply a fraud. Yes, you managed to improve Robo* in a fairly significant way. That doesn't change a thing, however....

Bob, can you keep that sort of wording to Rolf? With all kinds of trolls you engage in lengthy "intellectual" endeavors, both you and Rolfes polluting the forum and hijacking the threads. I don't think you have both qualifications and merit to attack in this manner Robert. And on topic, Houdini 1.5 parallelization _practically_ works better than Crafty's, with all 30 years of your "majestic" parallelization effort from Cray Blitz. Without talking of 3-400 Elo points gap...

Kai

[Albert Silver]

And on topic, Houdini 1.5 parallelization _practically_ works better than Crafty's

Data?

[Milos]

And on topic, Houdini 1.5 parallelization _practically_ works better than Crafty's

Data?

Well for any meaningful comparison it's also necessary for Bob to at least once present some data newer than those 30 years old from Cray Blitz...

[Laskos]

And on topic, Houdini 1.5 parallelization _practically_ works better than Crafty's

Data?

Don't have right now, but I did the following:
NPS and time to depth are useless.

1. I put them on 8 core machine at my dept. to play short TC games, and measured the strength. I compared this strength difference to 1 core difference, it was roughly by 20 +/- 10 (95% confidence) Elo points larger for Houdini 1.5 in 8 core mode (I repeat, relative difference 8-1 core, not the absolute one).

2. I selected 40 positions from WM test, solvable by both Houdini and Crafty on one core in less than a minute. Then performed 8 core test. Took the geometric average of the results (time used for each solution), and Houdini again came relatively better in 8 core mode (again, relatively 8-1 core Houdini to 8-1 core Crafty).

By the way, Rybka scales worse than both, after 6 years of another "majestic" efforts. I think some people have to appreciate the skills of R. Houdart.

Kai