SedatChess

[Ding-Bat]

Everest vs Don h would be an interesting none NNUE vs NNUE

[Sedat Canbaz]

Everest vs Don h would be an interesting none NNUE vs NNUE

Ok...I will run it unofficially... and soon as possible..

But very likely,
The results will be ended in favor for DON...

[Ding-Bat]

Thanks will still be a very interesting match

Everest vs Don h would be an interesting none NNUE vs NNUE

Ok...I will run it unofficially... and soon as possible..

But very likely,
The results will be ended in favor for DON...

[OrgZ]

Thanks.
This test is very interesting but am definitely sure nnue will dominate

Are you sure that NNUE will be dominant ?)

For example, according to new strength testings V:
DON Hercules 2 played even via Cfish 030820 NON-NNUE
This is not all.. I set 2m+1s for DON Hercules's book,
Where Lion 110423 book is played via Raubfish SC NNUE
Besides, I set 4m+2s for Lion 110423's opening book

And even under disadvantage test conditions,
DON NON-NNUE performed about 56 Elo better:

Code: Select all

1 DON Hercules 2 NON-NN  +16/-8/=26 58.00%   29.0/50
2 Lion 110423 NNUE       +8/-16/=26 42.00%   21.0/50

What I can say more,
Once more we noticed that DON fully deserves 'Hercules' name !)

Ok...that's all for now...I have many other new testings..
So..soon as possible, I hope to share all latest results


Greetings )

interesting book.

[Sedat Canbaz]

Thanks will still be a very interesting match

Everest vs Don h would be an interesting none NNUE vs NNUE

Ok...I will run it unofficially... and soon as possible..

But very likely,
The results will be ended in favor for DON...

Yes....

By the way,

Everest NON-NN will use 4m+1s,
DON NNUE will use 30sec+0.5 sec

[Sedat Canbaz]

Thanks.
This test is very interesting but am definitely sure nnue will dominate

Are you sure that NNUE will be dominant ?)

For example, according to new strength testings V:
DON Hercules 2 played even via Cfish 030820 NON-NNUE
This is not all.. I set 2m+1s for DON Hercules's book,
Where Lion 110423 book is played via Raubfish SC NNUE
Besides, I set 4m+2s for Lion 110423's opening book

And even under disadvantage test conditions,
DON NON-NNUE performed about 56 Elo better:

Code: Select all

1 DON Hercules 2 NON-NN  +16/-8/=26 58.00%   29.0/50
2 Lion 110423 NNUE       +8/-16/=26 42.00%   21.0/50

What I can say more,
Once more we noticed that DON fully deserves 'Hercules' name !)

Ok...that's all for now...I have many other new testings..
So..soon as possible, I hope to share all latest results


Greetings )

interesting book.

Yes...

By the way,
This time, I have good news,
Your Raid book is already proved as twice that
Its performance is better via NNUE assistance...

And soon as possible,
I hope to share the latest new results...

[Sedat Canbaz]

Everest vs Don h would be an interesting none NNUE vs NNUE

Your recent request is done in a few minutes,
And this time, no any winner, results equal:

Code: Select all

Score of DON Hercules 2 vs Everest NON-NN: 6 - 6 - 88 [0.500]
...      DON Hercules 2 playing White: 6 - 0 - 44  [0.560] 50
...      DON Hercules 2 playing Black: 0 - 6 - 44  [0.440] 50
...      White vs Black: 12 - 0 - 88  [0.560] 100
Elo difference: 0.0 +/- 23.6, LOS: 50.0 %, DrawRatio: 88.0 %
100 of 100 games finished.

As I started before (for anyone missed),
DON Hercules 2 via 30sec+0.5sec, Raubfish X48c3 SC
Everest NON-NN book via 4min+1sec, Cfish 030820


Will there be anything else?

[Ding-Bat]

Correct me if I'm wrong but is 300 games not better to get an average?
100 is short no?

Everest vs Don h would be an interesting none NNUE vs NNUE

Your recent request is done in a few minutes,
And this time, no any winner, results equal:

Code: Select all

Score of DON Hercules 2 vs Everest NON-NN: 6 - 6 - 88 [0.500]
...      DON Hercules 2 playing White: 6 - 0 - 44  [0.560] 50
...      DON Hercules 2 playing Black: 0 - 6 - 44  [0.440] 50
...      White vs Black: 12 - 0 - 88  [0.560] 100
Elo difference: 0.0 +/- 23.6, LOS: 50.0 %, DrawRatio: 88.0 %
100 of 100 games finished.

As I started before (for anyone missed),
DON Hercules 2 via 30sec+0.5sec, Raubfish X48c3 SC
Everest NON-NN book via 4min+1sec, Cfish 030820


Will there be anything else?

[Sedat Canbaz]

Correct me if I'm wrong but is 300 games not better to get an average?
100 is short no?

Everest vs Don h would be an interesting none NNUE vs NNUE

Your recent request is done in a few minutes,
And this time, no any winner, results equal:

Code: Select all

Score of DON Hercules 2 vs Everest NON-NN: 6 - 6 - 88 [0.500]
...      DON Hercules 2 playing White: 6 - 0 - 44  [0.560] 50
...      DON Hercules 2 playing Black: 0 - 6 - 44  [0.440] 50
...      White vs Black: 12 - 0 - 88  [0.560] 100
Elo difference: 0.0 +/- 23.6, LOS: 50.0 %, DrawRatio: 88.0 %
100 of 100 games finished.

As I started before (for anyone missed),
DON Hercules 2 via 30sec+0.5sec, Raubfish X48c3 SC
Everest NON-NN book via 4min+1sec, Cfish 030820


Will there be anything else?

100 games is enough... 300 games is too long for running direct matches
As main reason, usually Top trend books prefer strictly same/similar openings
Just in case of running 300...then may appear too many double games...

On other hand even if we run 1000 games...then I see no much reason...
Because the produced data will be no much useful for overall users
E.g the produced stats of Two books may lead to wrong conclusions...

But anyhow, I managed to run many games...sure with different cond.
But I assume will be more useful...later I hope to publish...

[OrgZ]

Everest vs Don h would be an interesting none NNUE vs NNUE

Your recent request is done in a few minutes,
And this time, no any winner, results equal:

Code: Select all

Score of DON Hercules 2 vs Everest NON-NN: 6 - 6 - 88 [0.500]
...      DON Hercules 2 playing White: 6 - 0 - 44  [0.560] 50
...      DON Hercules 2 playing Black: 0 - 6 - 44  [0.440] 50
...      White vs Black: 12 - 0 - 88  [0.560] 100
Elo difference: 0.0 +/- 23.6, LOS: 50.0 %, DrawRatio: 88.0 %
100 of 100 games finished.

As I started before (for anyone missed),
DON Hercules 2 via 30sec+0.5sec, Raubfish X48c3 SC
Everest NON-NN book via 4min+1sec, Cfish 030820


Will there be anything else?

You fear that using the same engine, the one with nnue ON might be too advantageous?
Have u tried cichess 040121? How is it compared to That version on cfish?

As far as on my tests with it, its got some venom. Its got the ability to get wins against latest stockfish devs.

Reason why am bringing this up Mr Sedat is to try to understand why you chose cfish 030820. As far as i know u look for an engine that can decline a maximum number of draws, so my question is have u tried cichess and see how it declines draws?.

You might find it interesting



Raid 3TR
I guess i'll see when the test is over Mr Sedat.