PRIVATE ProteusSFX-AI tests

[cubail]

ProteusSFX-AI_Mustafacanseven_2023.pgn
NOW LOSE YOU ENGINE SB ENGINE VINS
https://pixeldrain.com/u/MhQzh2ft

[cubail]

https://pixeldrain.com/u/X6NP1wQ6
https://pixeldrain.com/u/ZbBXp3Lj
YOU LOOK NAME CHANCE CLONE ENGINE

[AlexChess]

ProteusSFX-AI 231109 vs Hazard v.4.1 EXTtreme
Blitz 5+0 - Eman.bin book for the oppontent
Final results:

[AlexChess]

Next match is against Toto's Tactical 161223

[Martin_1969]

ProteusSFX-AI 231109 vs Hazard v.4.1 EXTtreme
Blitz 5+0 - Eman.bin book for the oppontent
Final results:

When I look at the +/-, errorbar all these engines are the same strength.
They are all just Stockfish derivatives and none are really stronger than their father.

[OrgZ]

ProteusSFX-AI 231109 vs Hazard v.4.1 EXTtreme
Blitz 5+0 - Eman.bin book for the oppontent
Final results:

When I look at the +/-, errorbar all these engines are the same strength.
They are all just Stockfish derivatives and none are really stronger than their father.

This nimrod morales is testing books not engines.
After every win his book gets updated by those wins won by other engine+ his. So ITS A BOOK FIGHT NOT ENGINE FIGHT.

Engine fight
You need Testsuite (reverse colours). All must be square no pre-learned exp no books then ull see results.

What he is doing is book testing nothing else. Chill

[AlexChess]

ProteusSFX-AI231109 vs TACTICAL-161223 FINAL RESULT: 4-5



Tactical has won 150.5-149.5 (+1/300)

@ORGZ True. It's a gauntlet: ProteusSFX-AI vs all Stockfish & SF derivatives, not a rating list like SPCC. Since not all engines include an opening book, I use generic Eman.bin for almost all opponents (if they support Polyglot) that is very good (created by Khalid Omar) or no book at all. For Light 12 Pro I have used TCEC unbalanced suite, repeating each position on both sides, as you suggest. But isn't useful for my purpose of creating 2 books with all won games between Stockfish and derivatives.

This is a PRIVATE engine test, to improve by books training them against all Playchess engines.

[AlexChess]

ProteusSFX-AI 231109 vs Hazard v.4.1 EXTtreme
Blitz 5+0 - Eman.bin book for the oppontent
Final results:

When I look at the +/-, errorbar all these engines are the same strength.
They are all just Stockfish derivatives and none are really stronger than their father.

+-14 ELO is the error margin after 300 games. I agree, no one util now is better than his father

[AlexChess]

Next match is against Incognito 5 (without Eman.bin, since it doesn't support Polyglot. Let's see if it scores better than Light 12 pro, by far the worse clone until now )

NOTE: On Playchess.com I have almost the worse hardware. I was curious to see my performance using the same minimal CPU for all opponents.

[AlexChess]

BANKSIAGUI ProteusSFX-AI 231224 GAUNTLET
300 5+0 blitz games against all Stockfish and SF derivatives on core i5 4 cores 2 Mn/s
231224 UPDATE (4200 GAMES)



Trust me, since I use these games to build my white and black books, results are REAL (I don't want to add noise to my books) . My formula? I add all won games by both engines + all 2700 ELO Super GMs won games + all Playchess games of last 2 years.

As expected, Blue Marlin and Crystal (that are very good on problems solving) are on bottom on blitz. Probably 2 GB Eman exp is too heavy for my hardware. Tactical is a surprise... It has won 5 games with white and lost 4 games with black (now all them added to my books against Incognito 5 )

Sorry, since the 260 won games (until now) are GOLD, I will send them only to my chess friends SARONA and dorsz