H5py

Important

Discoverer HPC provides public access to the h5py module included in TensorFlow and PyTorch installations.

Accessing h5py

To use h5py on Discoverer HPC you should stick to Intel Distribution for Python. To obtain access to h5py you should load either intel.universe.tensorflow or intel.universe.pytorch environment modules:

module load intel.universe.tensorflow

or

module load intel.universe.pytorch

Afterwards h5py can be imported by the Intel Python interpreter.

Checking the version

The easiest way to check the versions of the h5py and linked HDF5 library is to execute the following Slurm batch script:

#!/bin/bash
#
#SBATCH --partition=cn         # Partition name (ask the support team to clarify it)
#SBATCH --job-name=h5py
#SBATCH --time=00:01:00        # WallTime - one minute is more than enough here

#SBATCH --nodes           1    # May vary
#SBATCH --ntasks-per-node 1    # Must be 1
#SBATCH --cpus-per-task   1    # Must be 1

#SBATCH -o slurm.check_h5py_version.out        # STDOUT
#SBATCH -e slurm.check_h5py_version.err        # STDERR

module purge
module load intel.universe.pytorch

cd $SLURM_SUBMIT_DIR

python -c "import h5py;print('Linked HDF5 library:',h5py.version.hdf5_version)"
python -c "import h5py;print('H5py version:',h5py.version.version)"

To do that, store the script content into a file, for example /discofs/${USER}/check_h5py_version.sbatch and submit it as a job to the queue:

cd /discofs/${USER}/check_h5py_version.sbatch
sbatch check_h5py_version.sbatch

Then check the content of the file slurm.check_h5py_version.out to find out which verions of HDF5 and h5py are reported there.

Getting help

See Getting help