New J.A Compiles

[Jim Ablett]

@ supersharp77

A lot of interesting stuff there. I'll take look.

Jim.

[Jim Ablett]

https://github.com/visanalexandru/Fortuna
Fortuna 1.0 by Visan Alexandru

Fortuna 1.0 64 JA

Windows/Linux & Android profiled builds

https://drive.proton.me/urls/JJTQH37CAR#wtj7iGt1uzfB

Jim.

[Jim Ablett]

Hey Jim See You've Been Busy Indeed!! Thanks!!.... I've Got Quite A few Unknown Engines Perhaps You Can Assist On....Thanks I Advance....Incoming!!!

https://github.com/healeycodes/andoma/releases

"

My blog post: Building My Own Chess Engine


A chess engine which implements:

Alpha-beta pruning for move searching
Move ordering based off heuristics like captures and promotions
Tomasz Michniewski's Simplified Evaluation Function for board evaluation and piece-square tables
A slice of the Universal Chess Interface (UCI) to allow challenges via lichess.org
A command-line user interface

It uses Python 3.8 with Mypy type hints and unit + integration tests.

See Contributing to help out!

Install

pip install -r requirements.txt

I've already compiled this one (last May)

https://drive.proton.me/urls/DWYD2PKF0C#x4bKPzcdISnt

Jim.

[Jim Ablett]

@ supersharp77

Gofish consist of separate chess function modules for you to use to
write your own engine. Nothing to compile here.

Nothing to compile with Oracle. You need to add your own personal OpenApi key in the
python source.

Muzero is a neuro network training framework. Not an engine.

Leela is built using a compile script. If I were to compile it would be identical the original, so no point.

Converting Crafty to use UCI natively is possible but a lot of work. Possibly the easiest way to do it
would be to interface with the Senjo UCI adapter code > https://github.com/zd3nik/SenjoUCIAdapter
I will take a look.

Jim.

[Jim Ablett]

https://github.com/kelseyde/calvin-chess-engine
Calvin 4.3.0 by Dan Kelsey

Calvin 4.3.0 64 JA

Fast standalone Windows executable

https://drive.proton.me/urls/90AB2CV5MG#mXTp2053OKYq

Jim.

Update:

Calvin 5.1.0 64 JA

Standalone Windows executable

https://drive.proton.me/urls/BV053DG660#wqG9lBpvFpxj

Jim.

[Jim Ablett]

https://github.com/Yoshie2000/PlentyChess

PlentyChess 3.0.0 dev by Patrick Leonhardt

PlentyChess 3.0.0 dev 64 JA

Windows/Linux & Android clang profiled builds

https://drive.proton.me/urls/3RJZXSR7PC#74N4OMD7KD5S

Jim.

Update:

PlentyChess 3.0.0 64 JA

Windows/Linux profiled builds

https://drive.proton.me/urls/5BWT27P3B4#rpexAk1RpANX

Jim.

Update:

PlentyChess 4.0.1 64 JA

Windows/Linux profiled builds

https://drive.proton.me/urls/BV09HF36XW#ciZocedMAp7K

Jim.

[Joseph]

How about android version of PlentyChess 4.0.1?

[Jim Ablett]

How about android version of PlentyChess 4.0.1?

Hi Joseph,

I thought the author did one but I had a look and he hasn't.
I will compile an android version today.

Jim.

[Jouni]

Jim's PlentyChess compile is stunning 17% faster than official (bmi and windows) according to bench! But with more cores difference is much smaller. Same thing happened with Jim's Stockfish17 compile. JA compile was some 15% faster in bench. In github they didn't believe me at all, but said it's impossible. Weird.

[pohl4711]

avx2 and avx512 binaries of new Patricia dev. The avx2 binary is (strange) little bit faster on my machines than the avx512 binary... So try out, whats faster on your PC, before using avx512.

This new Patricia dev is a huge step forward: Inspired by my Cerberus-project (https://www.sp-cc.de/cerberus-project.htm), Patricia has now 2 neural-nets instead of one (a non-aggressive net was added for playing worse positions and the endgame). And in my tests, this new Patricia has the same Celo-strength as Patricia 4.0, but a clearly improved EAS-Score (+52000 points compared to Patricia 4.0 (!!!))

https://drive.proton.me/urls/82N6SJZCEG#wdJUCfNlq7OO