Closed jreadey closed 1 month ago
For linked datasets, use h5py to query chunk locations and store results to be used as needed for future queries. Add option to hsload to not initialize chunk locations.
chunk_iter requires hdf5 version 1.14.0 or higher and h5pyd 3.8.0 or higher.
hsload now supports a --fastline option that has chunk locations exctracted server side.
--fastline
For linked datasets, use h5py to query chunk locations and store results to be used as needed for future queries. Add option to hsload to not initialize chunk locations.