Hi all,
I have a 3D volumetric data which is organized as 1D array (for better
data management during the simulation). This is time-series data, so
basically I will have an array of
- 1D data (with size X*Y*Z)
- 3D data (with size (X,Y,Z))
when I save to output file.
My question is, to save this data to HDF5 file format. Should I keep using
1D linear array, or map to 3D dataset? Is there a specific HDF5 API designed
for dealing with 3D dataset?
Thanks,
Tuan
Hi Tuan,
Hi all,
I have a 3D volumetric data which is organized as 1D array (for better data management during the simulation). This is time-series data, so basically I will have an array of
- 1D data (with size X*Y*Z)
- 3D data (with size (X,Y,Z))
when I save to output file.My question is, to save this data to HDF5 file format. Should I keep using 1D linear array, or map to 3D dataset?
Generally speaking, you should make the data as self-describing as possible. In this case, that means you should map to a 3D dataset.
Is there a specific HDF5 API designed for dealing with 3D dataset?
The regular Dataset API (H5D* routines) is appropriate for this dataset.
Quincey
On Feb 7, 2011, at 8:52 PM, Hoang Trong Minh Tuan wrote:
