SPCC: Testrun of Stockfish 190531 finished

[pohl4711]

Testrun of Stockfish 190531 finished.

https://www.sp-cc.de


(Perhaps you have to clear your browsercache or reload the website)

[Raphexon]

Thank you for your continued testing.

Also I'm wondering if different net sizes behave differently on Allie than on Leela since Allie has a different (and faster) search.
Would it be possible to test Allie vs Leela on one of the smaller (distilled) nets too or a 10x128 net?
https://github.com/dkappe/leela-chess-w ... d-Networks

I wonder if Allie on a small net vs Leela on the same net would change their ELO relative to each other.
Since Allie's search is so different, does she want a bigger or a smaller net (than Lc0?)

[Raphexon]

And a big shame that SF doesn't seem to be progressing.

[pohl4711]

I wonder if Allie on a small net vs Leela on the same net would change their ELO relative to each other.
Since Allie's search is so different, does she want a bigger or a smaller net (than Lc0?)

On cccc Allie plays with the Leelenstein NNs, which have normal 20x256 size.

Perhaps, I will try a bigger 40x256 net with Allie later, but the main goal of my testings is to test lc0 and it's nets. And, because the hardware gets faster, I find bigger nets more interesting, than smaller nets. And mention, that one of my lc0 (or Allie)-testruns, with at least 700 games, takes 5 days. And the nets are changing quickly. And we have the 40xxx 20x256 nets and the 40x256 bootstrapped out of the 40xxx-nets. Both are in selflearning progress. So, there is a lot to test...

Stefan (SPCC)

[Raphexon]

I wonder if Allie on a small net vs Leela on the same net would change their ELO relative to each other.
Since Allie's search is so different, does she want a bigger or a smaller net (than Lc0?)

On cccc Allie plays with the Leelenstein NNs, which have normal 20x256 size.

Perhaps, I will try a bigger 40x256 net with Allie later, but the main goal of my testings is to test lc0 and it's nets. And, because the hardware gets faster, I find bigger nets more interesting, than smaller nets. And mention, that one of my lc0 (or Allie)-testruns, with at least 700 games, takes 5 days. And the nets are changing quickly. And we have the 40xxx 20x256 nets and the 40x256 bootstrapped out of the 40xxx-nets. Both are in selflearning progress. So, there is a lot to test...

Stefan (SPCC)

The 40x net would be great.

[stavros]

And a big shame that SF doesn't seem to be progressing.

if you think its easy at that level the proggress ,drop your ideas, go to fishcooking group,donate our hardware power etc etc.
in reality the shame is that there are many "just" users who enjoy the top free programs and dont give a @ about
the efforts of the teams who spent millions games,hours,electricity etc.. to improve one engine
its a big shame that you will never understand this.

[jorose]

And a big shame that SF doesn't seem to be progressing.

While single core progress hasn't been quite at the insane speed it has had in the passed, the arguably more important progress on 8 cores has been roughly 30 Elo over 6 months. That is fantastic progress if you ask me.