How does HDF5 engine work?

Hi All,

I'm curious about how the HDF5 engine works to insert or retrieve the data, because I already do a small benchmark to insert and retrieve data from HDF5, and the result is:
The file without any hierarchical and contains only datasets gives a lower performance on insertion and even retrieval rather than the file with hierarchical and the dataset is store in the group.

How can it be? I thought that it will be faster if the process directly retrieve from the dataset by giving the path name of that specific dataset, or my assumption was wrong?

Please help me and if you have any documentation or tutorial about this, I'll appreciate it.

Thank you.