NAMD¶

NAMD is a parallel molecular dynamics code designed for high-performance simulation of large biomolecular systems. NAMD's interface provides access to hybrid QM/MM simulations in an integrated, comprehensive, customizable, and easy-to-use suite. It has been developed by the collaboration of the Theoretical and Computational Biophysics Group (TCB) and the Parallel Programming Laboratory (PPL) at the University of Illinois Urbana–Champaign.

Availability and Supported Architectures¶

NAMD runs at NERSC are currently supported on CPU and GPU nodes.

Application Information, Documentation and Support¶

The official NAMD user-guide is available at NAMD User Guide. NAMD has a large user base and good user support. Question related to using NAMD can be posted to the NAMD Mailing List. The forum also contains an archive of all past mailing list messages, which can be useful in resolving some of the common user issues.

Using NAMD at NERSC¶

NAMD is now supported on Perlmutter machines using containers. Please note that NAMD modules are deprecated as of 20^th April 2024, and will not be updated.

There are two different containers of NAMD available on Perlmutter:

perlmutter$ shifterimg images | grep 'namd'
perlmutter docker     READY    91c54bf1e5   2024-03-21T13:36:27 nersc/namd:3.0.b5
perlmutter docker     READY    cd62d5f2a4   2024-03-21T13:09:34 nvcr.io/hpc/namd:3.0-beta5

The nersc/namd container is built to run NAMD simulations on the CPU nodes. The nvcr.io/hpc/namd container comes built from the NVIDIA container repository. Both these containers are compiled to NAMD version 3.0 beta 5 release. The nersc/namd container image is built with MPICH support for Charm version 7.0.0. This container supports multi-node and mult-thread runs. The nvcr.io/hpc/namd conatiner, designed to run on GPUs DOES NOT have multi-node support.
Users are encouraged to test the container images for scaling performance before submitting production runs.

Using NAMD on Perlmutter¶

As stated NAMD can be run on both, CPU and GPU nodes of Perlmutter. The following are the two example scripts that can be used to submit a batch job to either of the nodes.

#!/bin/bash
#SBATCH --image docker:nersc/namd:3.0.b5
#SBATCH -C cpu
#SBATCH -t 00:20:00
#SBATCH -J NAMD_CPU
#SBATCH -o NAMD_CPU.o%j
#SBATCH -A <your_nersc_project>
#SBATCH -N 16
#SBATCH --ntasks-per-node=128
#SBATCH -q regular

exe=namd3
input= "+setcpuaffinity ++ppn 4 <please change this as per your requirements>"

export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK
export OMP_PROC_BIND=spread
export OMP_PLACES=threads

command="srun --cpu-bind=cores --module mpich shifter $exe $input ./<name of the input file>" 

echo $command

$command 

The above script launches a 16 CPU node job running with 128 tasks (equal to the number of cores available on 1 CPU node).

#!/bin/bash
#SBATCH --image nvcr.io/hpc/namd:3.0-beta5
#SBATCH -C gpu
#SBATCH -t 00:20:00
#SBATCH -J NAMD_GPU
#SBATCH -o NAMD_GPU.o%j
#SBATCH -A <your_nersc_project>
#SBATCH -N 1
#SBATCH -c 16
#SBATCH --ntasks-per-node=4
#SBATCH --gpus-per-task=1
#SBATCH --gpu-bind=none
#SBATCH -q regular

exe=namd3
input="+setcpuaffinity ++ppn 16 +devices 0,1,2,3 <please change this as per your requirements>"

export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK
export OMP_PROC_BIND=spread
export OMP_PLACES=threads

export CUDA_VISIBLE_DEVICES=0,1,2,3
export CRAY_CUDA_MPS=1

command="srun --cpu-bind=cores --gpu-bind=none --module mpich,gpu shifter $exe $input ./<name of the input file>" 

echo $command

$command 

The example above uses 1 GPU node on Perlmutter, which has 4 GPUs each. When changing the number of nodes, please modify the line SBATCH -N 1 to whatever number of nodes you want to run your problem with. Additionally, please change Please change line 'input=' in accordance to NAMD use guide as suitable for your problem.

Further details on using docker containers at NERSC with shifter can be found at shifter

Related Applications¶

• LAMMPS

User Contributed Information¶