Latest Update: 08.04.2023


Dear Chess Friends,


And as usual, I'm pleased to announce that,

Here is another new GUI/Engine competition!


1st of all,

The idea is born after both tournaments: CS VI / Candidate

And later, I realized to see what happens via more games


On other hand, really it is as mystery which GUI produces

The right and more accurate results.. frankly no any idea, 

But it's clear that Cfish 030820 or CF EXT 291020 suffers...!

Sure probably the programmers will explain better than Me


But anyhow..for example according to my experience,

After comparing... the results are really different..sure

We can not expect exactly same/identical points...but

It seems, there are serious influences. factors etc.

Where as we see the Elo differences are about 10 Elo

And I don't think that due to Error margin...because

Especially 2 core test contains 1238 games (per GUI)

Btw, for better cond. each opening is played as twice!

I mean, here the openings can not be as the reason..


Meanwhile, without to not mention this I can not as well )

I know too that any engine performance is highly depends

On our used openings..but I was not aware so much that

GUIs can play some role over the engines strength too !


Note also that,

     For both GUIs are installed different engine exe copies

     (Both GUIs are not used to play from same engine exe)


Ok... here are the latest played GUI / Engine results:


CuteChess v5.11.1, 4 Core/Game, Unique Openings,

512 MB Hash-table, Concurrent games: 18


30 Elo difference (in favor for Cfish 030820)

                   

1   Cfish 030820      +74/-48/=178 54.33%  163.0/300

2   CF EXT 291120  +48/-74/=178 45.67%  137.0/300


Draws: 60% / White Wins: 29% / Black Wins: 11%


---------------------------------------VS---------------------------------------


Banksia 0.58, 4 Core/Game, Unique Openings,

512 MB Hash-table, Concurrent games: 18


22 Elo difference (in favor for Cfish 030820)

                                    

1   Cfish 030820      +67/-48/=185 53.17%  159.5/300

2   CF EXT 291120  +48/-67/=185 46.83%  140.5/300


Draws: 62% / White Wins: 26% / Black Wins: 12%


================================================

CuteChess v5.11.1, 2 Core/Game, Balsa Openings,

128 MB Hash-table, Concurrent games: 36


22 Elo difference (in favor for Cfish 030820)

                                 

1   Cfish 030820      +200/-123/=915 53.11%  657.5/1238

2   CF EXT 291120  +123/-200/=915 46.89%  580.5/1238


Draws: 74% / White Wins: 16% / Black Wins: 10%


---------------------------------------VS---------------------------------------


Banksia 0.58, 2 Core/Game, Balsa Openings,

128 MB Hash-table, Concurrent games: 36


12 Elo difference (in favor for Cfish 030820)                

                   

1   Cfish 030820      +200/-123/=915 53.11%  657.5/1238

2   CF EXT 291120  +123/-200/=915 46.89%  580.5/1238


Draws: 75% / White Wins: 16% / Black Wins: 9%


==================================================
More Details, I used Babksia 0.58  (18 Jan 2023) / Cutechess Qt v5.1.11 (9 Aug 2020)Cfish 030820 is Popcnt / CF EXT is SSE41/ tested as NON-NNUEPlus both are played with move overhead 100 and via Contempt 40  And Contempt 40 is played an important role about 75% draw-ratio!  Sure these stats are via Balsa openings, but with Unique: even 62%  That means also no any other engine to be strong..and in same time  To produce so low draw numbers...in other words, simply just great!  But I can't say same words about why there is minus plus 10 Elo diff...  Anyhow, I think here it's related and the reason is GUI, but which one  Not sure....but as I mentioned above...it seems to be as mystery...!)• Download all the played games (in PGN): GUI / Engine Testings IV

 Well, that's all for now...and many thanks for your interest...)




Best,Sedat