@gotwic said [^](/forum/redirect/post/w7U1r45R)
> ok thank you, that clarifies my earlier obervations about lichess handles being credited with fishnet analysis at least :)
do you think its worthed to wait a few minutes for the 2 sec game analysis with around 6 m nodes ?
In some positions 1M is complete trash due to low depth, but in most cases it's more than enough so probably not.
@gotwic said [^](/forum/redirect/post/XJKj2CUg)
> In some cases 1M is complete trash due to low depth, but in most cases it's more than enough so probably not.
i wish there would be something in between like 1 sec for example
Yeah we need to clamor for an additional nodes per move slider. Then all will be peachy.
edit:
looks like the idea was already brought up in https://github.com/lichess-org/lila/issues/22127
Yeah we need to clamor for an additional nodes per move slider. Then all will be peachy.
edit:
looks like the idea was already brought up in https://github.com/lichess-org/lila/issues/22127
i found a game with published local analysis for some reason 
@gotwic said [^](/forum/redirect/post/kGVRct40)
> Yeah we need to clamor for an additional nodes per move slider. Then all will be peachy.
>
> edit:
>
> looks like the idea was already brought up in https://github.com/lichess-org/lila/issues/22127
i found a game with published local analysis for some reason 
A few notes:
Upload analysis IS possible, but only for studies where you're a contributor or owner.
Standard would be 1,000,000 moves with Stockfish 94MB using 1 thread. The more threads you use, the more nodes you need. This is due to inefficiencies in search space partitioning that are only significant in smaller searches (under a few million nodes). Even when significant, the effect is rather small compared to the benefit of multi-threading. Lichess has measured this effect and does the number crunching for you. It adjusts the target nodes based on your threads setting so you can be confident that standard is the same strength as Fishnet. And more threads is still much faster, up until the number of physical cores on your CPU.
All other browser engines are inferior to Stockfish 94MB. Even the 1MB light version needs to crunch 7-8x the number of nodes to achieve a similar analysis strength as 94MB. The weakness rises significantly with age. Older engines such as Stockfish 11 HCE might take 100x-1000x as long to very roughly match standard, so it is best to use timed evaluations if you're forced to use one of the old guys.
Another interesting thing to consider is that Fishnet evaluations from 5-6 years ago are multiple orders of magnitude weaker than the Fishnet evals of today. So "standard" does not mean the strength of all fishnet evals. It means exactly the strength of a Stockfish 19 Fishnet evaluation.
@gamblej said [^](/forum/redirect/post/cUbF5keZ)
> A few notes:
>
> Upload analysis IS possible, but only for studies where you're a contributor or owner.
>
> Standard would be 1,000,000 moves with Stockfish 94MB using 1 thread. The more threads you use, the more nodes you need. This is due to inefficiencies in search space partitioning that are only significant in smaller searches (under a few million nodes). Even when significant, the effect is rather...
what does the amount of ram chanche ?, i once tested analysis board stockfish with min and max ram and got the same calculation speed
Not much on smaller searches.
@gamblej said [^](/forum/redirect/post/2zpSAvc5)
> Not much on smaller searches.
how would you define a small search