MX Linux Forum

I found this in wikipedia:
"NNUE runs efficiently on central processing units without a requirement for a graphics processing unit (GPU)."

NNUE is a neural net evaluation function used by many game programs to evaluate positions at the end of a calculated variation.

It is now used in stockfish which is in the MX repositories. I think stockfish 12 is available. It uses the NNUE evaluation function, though it is a option, and you can revert to the old stockfish hand crafted evaluation function.

The fully neural net programs like Lc0 are close in strength to stockfish, maybe a bit higher or lower in elo rating.
But stockfish 12 is plenty strong for me (of course crushes me a poor human) but also strong enough for deep analysis. It is currently rated at >3600 elo, so strong enough to crush Magnus! This is strong enough for me. As for a machine, any fast multi core Ryzen is awesome. I run stockfish with 10 threads on my 12 thread (6 core) machine, leaving two threads for MX and firefox.

Perhaps I forgot to say that my interest is only focused on engine competition; I'd like to see what you and others say about lc0 vs stockfish =

Here I see no such thing, even Koivisto an engine that is ranked #8 beats lc0 easily and is always under time pressure.

If you test it in inter-engine competition you will confirm what I am saying -- unless someone is using some built-in gadgetry which is what some seem to indicate ...

Thanks for your comentary

https://www.amd.com/en/products/profess ... ro-wx-9100
My dream card right there.
eats LC0 for breakfast. Of course you will need complimentary hardware to match.

Does not your program you want to run have a "system requirements" document? Why not consult with LCO for the expected system requirements and then meet them in your hardware? Hell, I don't even know what you are talking about and I don't see this as an MX issue.

Seaken64

As i said in my post ... This is what someone knowledgeable in lc0 says about requirements:

"Without GPU, Lc0 is very slow as CPU is not very well suited to evaluate neural networks.

On CPU only, Lc0 is indeed weaker than other top chess engines, even though it's still much stronger than any of human players."

But I'm not an expert in any of these ...

How should I interpret this, as a hardware issue only or a software issue. Because if it is only a hardware issue, as to my poor understanding it is, then I need another machine -- what do you recommend, which one...?.

Thanks.

You can play for free online against powerful (elo is adjustable) chess engines.
https://www.shredderchess.com/online/pl ... nline.html
https://www.sparkchess.co
https://www.365chess.com/play_computer_online.phpm/

Stockfish 15.1 in the MX (test) repos also includes a NNUE compiled into it at build time.

Wikipedia says:

Stockfish has consistently ranked first or near the top of most chess-engine rating lists and, as of February 2023, is the strongest CPU chess engine in the world. Its estimated Elo rating is over 3500.[3] It has won the Top Chess Engine Championship 13 times and the Chess.com Computer Chess Championship 19 times.

It has been ported to browser code and can run in a web page, too.

I had not seen this last post...

The problem with the chess engines that one sometimes looks to install is that not only does a player use "the best program" whether installed on their PC, or online, it's not about playing against the engines but installing them on the operating system so that the engines can play against each other and watch their ...

It's not about program vs. human but program vs. program!

That's why 'we' are looking for a better machine/hardware ...

-- I also think (suggestion) that maybe it would be good to have at least the top 5 chess engines available/updated in MXLinux,

"Open source"

http://ccrl.chessdom.com/ccrl/4040/

Greetings.



Translated with www.DeepL.com/Translator (free version)

Can't you ask the other players on the chess forums what hardware they would recommend?

I did it, that was some time ago, I received different answers and I did not understand very well the issue so I abandoned the matter, I did not persist because at that time I was motivated by a single type of engine that did not run well on my machine but nowadays that engine is not of much interest,

-- You know what would always be a very good idea? that MXLinux would include the best engines in the system, that would always be of great value.

Thanks