...
To verify GPU access, use either the rocm-smi (AMD GPU servers) or nvidia-smi (NVIDIA GPU servers).Two Python scripts are located in /stor/scratch/
GPU
...
- Tensor Flow
- python3 /stor/scratch/GPU_info/tensorflow_example.py
- PyTorch
- python3 /stor/scratch/GPU_info/pytorch_example.py (test script not yet working on AMD servers)
- python3 /stor/scratch/GPU_info/pytorch_example.py (test script not yet working on AMD servers)
If GPUs are available and accessible, the output generated will indicate they are being used.
GPU-enabled software
See server-specific pages for how to run TensorFlow, PyTorch and AlphaFold and GROMACS.
Sharing resources
Since there's no batch system on BRCF POD compute servers, it is important for users to monitor their resource usage and that of other users in order to share resources appropriately.
...