Storion Wiki

To setup your envirment of razor128 On master node

module load scripts/python/3.12_2025v1

python -m venv ~/venvs/torch

source ~/venvs/torch/bin/activate

pip install --pre torch --index-url https://download.pytorch.org/whl/nightly/cu128

pip install sevenn

To test

Download test and upload it to razor128

source ~/venvs/torch/bin/activate

cd test_gpu

sbatch test_gpu.sh

cat gpu_test.out

It should show

Python 3.12 is now available.
Device: cuda | CPUs: 8
Atoms(symbols='LiF', pbc=True, cell=[[0.0, 2.0135, 2.0135], [2.0135, 0.0, 2.0135], [2.0135, 2.0135, 0.0]])
Lattice constant: 2.848 Å
cueq
False
flash
False
Initial energy: -9.6619 eV
      Step     Time          Energy          fmax
FIRE:    0 20:00:21       -9.661907        0.000000
Final energy: -9.6619 eV

This efficiency of paralleling for uMLIP