Last Update: 23.09.2022

Hello Chess Friends,

As usually, I am pleased to announce that,

New NON-NNUE Bullet (1m+1s) tour is organized!

Note that,

The current competition is already ended..the reason

Once more we noticed no much progress since 2020...

It seems mostly programmers work only over NNUE

Sure I am referring about creators of engines 3500+

Meanwhile,

Why I like testing the engines as NON - NNUE too? in short:

The eng matches are more exciting and less draws for sure!)

That means also that more mistakes, more blunder moves...

Like Human-game play .Btw, just imagine...if all the played

Moves would be 'perfect' then all games can be as draw...

But this is also true that .just 2 years ago....all engines were

Without any EvalFile, and since NNUE appeared.. day by day,

Many engines are started to be released with Evalfile...sure

Mostly testers (including myself) are started to test as NNUE

Personally I test them mainly due to strength, but not always,

Even mostly of my tours played as NON - NNUE, because

I am BIG fun of CLASSİC chess and as I stated earlier:

Via huge assistance (gaining from previous played data):

Any FOOL can manage to play stronger chess...right ?

As other reason and to be more clear,

What about the draw-ratio via NNUE ON? e.g in my recent

Organized competition, 3750+ NNUE engines are produced

98% draws, for Results, btw, if via deep strong opening lines:

Probably over than 99% are expecting to be ended as draws!

Mainly via fast Bullet or vs weaker engines or weak openings...

We have some chances to drop the overall draw percentage...

In other words, what I am trying to explain that sometimes

It's not so bad idea the engines to be tested under Classic

Conditions as well..if nothing else just pure, original chess!

Be aware of that too (as participants, as far as possible):

New engine versions that never played before are picked...

Shachchess 21.1 (not v24) is played, due to good: Results

But here, Shachchess 21.1 performed below than expected..

Actually same words are going for many other engines too...

And I've realized to test again some of previous Winners too

E.g Cfish 11 is the same engine vers. (x86_x64) for Results

And SF200620MZ engine is the same one too, for Results

Plus Cfish 030820 as well..sorry that I can't publish all of the

Results....because too many.links, where Cfish is as Winner

Anyhow via this method (including older versions), I think that

It's much more clear regarding the Elo improvements...right?

And let me say please some more notes over Cfish 030820,

This chess engine is the strongest one!! not only under Blitz..

Plus under Bullet cond...again and again is Number One...!!

Undefeated Champion ..managed to be 6-times as Champ!

No more words to say exception just: Unbelievable !!

I have to admit also that,

StockfishMZ is really quite strong too, it's true that it is not so stable...

E.g suffers (often crashes) as DarkSister under Cutechess GUI, but

Under Arena GUI: both mentioned engines worked flawlessly... but

I can't say same words for Stockfish+190122 (by Kranium) it's buggy..

E.g Stockfish+ crashed many times under Arena ...no any idea why..

And I wonder too. If the mother is Stockfish, then how some SF based

Ones can be so buggy..? because the real 'mother' is pretty stable.. !?

Btw, BlueMarlin 15.3a, plus ProteusSF-Piranha 040922 suffers too e.g:

Theirs performances were much lower than expected...strange really..

It's like, without navigation: you can't find your home from 100 meters)

On other hand, the good news is that:

I did not see any crash by the rest participated engines, well-done!

Ok....and here are the Final Standings:

The Champion: Cfish 030820 - My Congratulations to Cfish, SF Team!

# PLAYER : RATING POINTS GAMES (%)

1 Cfish 030820 : 3572.3 349.0 640 54.5%

2 StockfishMZ 170522 : 3569.0 346.0 640 54.1%

3 Polyfish 260822 : 3565.9 343.0 640 53.6%

4 SF-PB 130722 : 3564.3 341.5 640 53.4%

5 Stockfish 170922 : 3559.5 337.0 640 52.7%

6 SF200620MZ : 3559.0 336.5 640 52.6%

7 Kayra 1.6 : 3555.8 333.5 640 52.1%

8 RedFish 220921 : 3552.1 330.0 640 51.6%

9 SF Polyglot 190922 : 3551.1 329.0 640 51.4%

10 Cfish 11 : 3550.0 328.0 640 51.3%

11 Stockfish XIr5 : 3548.4 326.5 640 51.0%

12 Fisherov 0.98m : 3547.9 326.0 640 50.9%

13 DarkSister 1.9a : 3544.7 323.0 640 50.5%

14 Corchess 170922 : 3541.0 319.5 640 49.9%

15 BrainLearn 18 : 3539.4 318.0 640 49.7%

16 Zeus 11 : 3536.8 315.5 640 49.3%

17 ShashChess 21.1 : 3520.4 300.0 640 46.9%

18 Crystal 4.1 : 3518.8 298.5 640 46.6%

19 ProteusSF v006 : 3509.8 290.0 640 45.3%

20 Swordfish 15.3a : 3504.5 285.0 640 44.5%

21 AlphaZero : 3460.4 244.5 640 38.2%

• Calculated by Ordo program / Cfish 11 engine is fixed to 3550 Eio

More Details,

• All of the engines are 64-bit, BMI2 versions used where available

• Due to some engine crashes (StockfishMZ...), Arena GUI used too

• Note also that each chess engine is played with default Contempt

• Overall draw percentage (based on 6720 games) not so high 79%

• For anyone interested, this great engine is available: Cfish 030820

• Download all the played tournament games in PGN: NON-NNUE V