You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For whatever reason .compute() is currently used at the end of convert_to_color to get the Sv data loaded into memory. This assumes that the dataset is loaded as a dask array, rather than a numpy array (which would actually results the plotting to error out, since .compute() does not exist for numpy arrays.
Removing .compute() works for when we load a xr.Dataset with chunks=None so that the content are numpy arrays. Need to test if without .compute() the function would still work normally.
The bigger picture though, behind this issue, is to review the current handling of input types (and chunking schemes?) in Echoshader, and see how that jives with performance when the data is large or small.
The text was updated successfully, but these errors were encountered:
For whatever reason
.compute()
is currently used at the end ofconvert_to_color
to get the Sv data loaded into memory. This assumes that the dataset is loaded as a dask array, rather than a numpy array (which would actually results the plotting to error out, since.compute()
does not exist for numpy arrays.Removing
.compute()
works for when we load axr.Dataset
withchunks=None
so that the content are numpy arrays. Need to test if without.compute()
the function would still work normally.The bigger picture though, behind this issue, is to review the current handling of input types (and chunking schemes?) in Echoshader, and see how that jives with performance when the data is large or small.
The text was updated successfully, but these errors were encountered: