VMD

VMD is pre-installed on this GPU machine. For now, you can only use VMD as a command-line (cli) tool. The graphic user interface (GUI) of VMD does not work with X2Go because the GLX problem. See this link for more information. If you want to use VMD GUI, you need to open VMD locally on the GPU workstation.

VMD is available at /data/software/vmd/bin

LAMMPS

Available Versions

Run Simulation

Just copy code below to a new file, e.g., run_lmp_gpu.sh, and then make this file executable by chmod +x run_lmp_gpu.sh . To run the simulation, simply type ./run_lmp_gpu.sh

#!/bin/bash 

# change the path to the appropriate one for your desirbed LAMMPS version
export LMP_GPU_PATH=/data/shared/ce-kaihangs_puser/lammps/lammps-gpu/lammps-2Aug2023/src/ 

# run program
# lammps.in is the input file
# pick ONE of the following options that suits your purpose

# Option 1
# running on 1 GPU (default)  
$LMP_GPU_PATH/lmp_gpu -sf gpu -in in.lammps

# Option 2
# because not all MD styles are supported by GPU version, we can use CPU to run as well
# running with CPU only, -np specify how many CPUs to use (in this case 16 cores)
#mpirun -np 16 ${LMP_GPU_PATH}/lmp_gpu -in lammps.in

gRASPA (under development)

Available Versions

Run Simulation

Just copy code below to a new file, e.g., run_graspa, and then make this file executable by chmod +x run_graspa . To run the simulation, simply type ./run_graspa . Note that all simulation input files need to be under the same directory.

#!/bin/bash

# set gRASPA path
export gRASPA_DIR=/data/shared/ce-kaihangs_puser/gRASPA/gRASPA-main/src_clean

# set CUDA device. There are two RTX 4090, set to "0" or "1". 
# You can use "nvidia-smi" command to check available devices
export CUDA_VISIBLE_DEVICES="1"

# run simulation and store output to output.txt file
$gRASPA_DIR/nvc_main.x > output.txt &

RASPA2

Available Versions