Pronto User Documentation

local testing

To build the documentation for local testing:

python3 -mvenv env
. env/bin/activate
pip install -r requirements.txt
mkdocs build
python3 -m http.server --directory site

deploy

Openshift will automatically build this when pushed for production.

update scenarios

hardware update

Use this to generate a new sinfo dump and save to data/sinfo.txt

sinfo --noheader -o '%n|%P|%X|%Y|%Z|%m|%f|%G' | sort

If adding a new GPU node, update GPU RAM and compute capability in main.py (search for base_gpu_info)

Manually add processor information to data/procs.txt

partition update

Copy the partition-related lines from slurm.conf into data/slurm.conf

ml-gpu update

Use this to gather the information needed for docs/machine_learning/preinstalled_software.md

singularity exec --nv /opt/rit/singularity/images/ml-gpu/VERSION/ml-gpu.sif /usr/local/cuda/bin/nvcc --version | sed -n 's/^.*release \([0-9]\+\.[0-9]\+\).*$/\1/p'
singularity exec --nv /opt/rit/singularity/images/ml-gpu/VERSION/ml-gpu.sif python3 --version
singularity exec --nv /opt/rit/singularity/images/ml-gpu/VERSION/ml-gpu.sif R --version | head -n 1
singularity exec --nv /opt/rit/singularity/images/ml-gpu/VERSION/ml-gpu.sif python3 -c 'import torch; print(torch.__version__)'
singularity exec --nv /opt/rit/singularity/images/ml-gpu/VERSION/ml-gpu.sif python3 -c 'import tensorflow; print(tensorflow.__version__)'
singularity exec --nv /opt/rit/singularity/images/ml-gpu/VERSION/ml-gpu.sif python3 -c 'import keras; print(keras.__version__)'
singularity exec --nv /opt/rit/singularity/images/ml-gpu/VERSION/ml-gpu.sif python3 -m pip freeze

Update recommended_mlgpu_version in main.py