Chunked HDF5 DataSet and slabsize -
We are evaluating HDF5 performance about chunked datasets. Especially we try to understand that is it possible to read in different parts and how this performance is affected? Like we have a dataset with a size of 10 digits, want to read a dataset with 100 data and 23 to 48 values. Will there be a big loss of performance?
Many thanks!
I do not know how to answer your question in particular, but I'm going to give you 1024 (or Some of the high power two). I do not know the interior of HDF 5, but with my knowledge of filesystem, and we had 1024 right from some benchmark.
Comments
Post a Comment