Stockfish at NCM: finally progress

Jouni · Post by **Jouni** » Mon Nov 22, 2021 1:01 pm

https://nextchessmove.com/dev-builds

+15 with optimism

Ozymandias · Post by **Ozymandias** » Mon Nov 22, 2021 1:14 pm

Actually, +6.67 to 20210915-1550. Optimism is masking some recent regression.

ernest · Post by **ernest** » Tue Nov 23, 2021 3:53 am

Indeed, Jouni !
To be followed...

Let's see what SPCC finds...

Jouni · Post by **Jouni** » Tue Nov 23, 2021 5:00 pm

Note, that 3 runs were within 1.21 points! My short test gave +11 to SF 14.1 against "weak" engines.

ernest · Post by **ernest** » Sat Nov 27, 2021 7:12 pm

ernest wrote: ↑Tue Nov 23, 2021 3:53 am
Let's see what SPCC finds...

forum3/viewtopic.php?f=6&t=78737

Jouni · Post by **Jouni** » Sun Nov 28, 2021 11:22 am

But new net nn-3678835b1d3d.nnue seems to be -10

. SF has reached it's maximum?

pohl4711 · Post by **pohl4711** » Sun Nov 28, 2021 2:51 pm

Jouni wrote: ↑Sun Nov 28, 2021 11:22 am But new net nn-3678835b1d3d.nnue seems to be -10 . SF has reached it's maximum?

IMHO testing vs. a so much weaker engine makes no sense. Stockfish wants to make progress vs strong opponents. Who cares about Stockfish scoring +440 or +430 Elo vs. SF 7?!? Meaningless.

We have Regression-tests vs. SF14 with classical openings in the SF-Framework and regression-tests using UHO openings vs SF 14.1 on my website:
https://www.sp-cc.de/stockfish-regression.htm

Personally, I dont need NCM...

Stockfish at NCM: finally progress

Stockfish at NCM: finally progress

Re: Stockfish at NCM: finally progress

Re: Stockfish at NCM: finally progress

Re: Stockfish at NCM: finally progress

Re: Stockfish at NCM: finally progress

Re: Stockfish at NCM: finally progress

Re: Stockfish at NCM: finally progress