Test High-Res MIP #73
Conversation
|
Oh yes. I activated the 'full' dynamic chunking for the first time, and look at this beautiful test dataset @rsignell: 
I am quite happy with this hehe. |
|
Trying the other datasets now as PR versions (which only loads two timesteps), will then move the iids to the main iid list and merge. Onwards! |
|
They all worked, yay. Lets get this on the road! |
|
Woohoo! I'm glad this worked! Is the final dataset going to be chunked 1 in time though? |
|
No this is a downside of the pruning we do here...the dynamic chunking function only sees 2 files, and thus the final chunking will be different on the final product... |
|
@jbusecke where do we stand on this Hi-Res MIP dataset? |
|
We were running into a pretty gnarly error that (in my intuition is caused by very large datasets with 'orthogonal' rechunking (going from time only to 'chunked all over the place') - read exactly these HighResMIP datasets). I am currently pushing hard on getting an MRE for this and hopefully that can serve as fuel to fix that bug upstream. Sorry for the delay again... |
|
Thanks for keeping this alive, @jbusecke ! |
#72