Glotzer lab members

Members of the Glotzer lab can install precompiled packages. Follow the instructions in on the installation page to install the latest stable release using the following channel paths and module versions:

HPC resourceChannel locationModule versions
UMich Great Lakesfile://nfs/turbo/glotzer/software/condamodule load gcc/10.3.0 openmpi/4.1.6 cuda/12.3.0
Purdue ANVILfile://anvil/projects/x-dmr140129/software/condamodule load gcc/11.2.0 openmpi/4.1.6
NCSA Deltafile://projects/bbgw/software/condamodule load openmpi/4.1.6 cuda/12.3.0
OLCF Frontierfile://ccs/proj/mat110/software/frontier/condamodule load PrgEnv-gnu rocm/5.4.3; module unload darshan-runtime
OLCF Andesfile://ccs/proj/mat110/software/andes/condamodule load gcc/10.3.0 openmpi/4.1.2


Individual users should install conda compatible environments in their home directory on Frontier. Importing Python packages from this environment will be very slow with large node count jobs. To improve performance, generate a tar file from the environment and store it on Orion.

$ tar --directory {{ environment-path }} -cf ${MEMBERWORK}/mat110/conda-env.tar .

Important: Repeat this step after you install or update packages with {{ package-manager }}.

Note: Collaborative projects may maintain a single copy of the software in the shared project directory: /ccs/proj/mat110/software/frontier/{{ subproject-name }}.

Collaborative projects may also utilize a single cached conda-env.tar:

$ tar --directory /ccs/proj/mat110/software/frontier/{{ subproject-name }} \
  -cf ${PROJWORK}/mat110/software/{{ subproject-name} }/conda-env.tar .

Use the following lines in your job scripts (or interactively with salloc) to load the environment into NVME and execute software from there:

#SBATCH -C nvme

module load PrgEnv-gnu rocm/5.4.3
module unload darshan-runtime

export CONDA_ENV_ROOT=/mnt/bb/${USER}/conda-env
srun --nodes=$SLURM_JOB_NUM_NODES --ntasks-per-node 1 mkdir ${CONDA_ENV_ROOT}
srun --nodes=$SLURM_JOB_NUM_NODES --ntasks-per-node 1 tar --directory ${CONDA_ENV_ROOT} -xpf \
#     ${PROJWORK}/mat110/software/{{ subproject-name }}/conda-env.tar # For use with shared projects.


srun {srun options} command arguments

Note: The above script has been tested on environments with all packages installed into base. You may need to set additional environment variables or source activation scripts to activate conda environments within this directory.

Development of glotzerlab-software is led by the Glotzer Group at the University of Michigan.

Copyright © 2024 The Regents of the University of Michigan.