DL_MESO

Overview

DL_MESO is a general purpose mesoscale simulation package which supports both Lattice Boltzmann Equation (LBE) and Dissipative Particle Dynamics (DPD) methods.

Version 2.8 is now available. It was compiled with Intel oneAPI 2024.2.0. It has additional DPD executables which can be run on GPUs. All utilities are also available.

Version 2.7.x is available. It was compiled with Intel 17 compilers (-msse4.2 -axCORE-AVX512,CORE-AVX2,AVX) . A version of DPD has been compiled with FFTW3 (see below). Utilities were compiled with the default GNU 4.8.5 compiler.

Version 2.6 is available and has been compiled with Intel 17 compilers.

The java interface is not available in any version.

Restrictions on use

Whilst the software is free for academic usage there are limitations within the DL_MESO license agreement which must be strictly adhered to by users. All users who wish to use the software must request access to the ‘dlmeso’ unix group. A copy of the full license is also available on the CSF in $dlmeso_home/$dlmeso_ver/LICENCE. Important points to note are:

  • No industrially-funded work must be undertaken using the software. See clauses 2.1.3 and 2.2 of the license.
  • The software is only available to Staff and Students of the University of Manchester. Users are reminded that they must not share their password with anyone, or allow anyone else to utlise their account.
  • Citation of the software must appear in any published work. See clause 4.2 for the required text.

There is no access to the source code on the CSF.

To get access to the software please contact us to confirm that your work will meet the above T&Cs.

Set up procedure

Once you have been added to the unix group please load ONE of the following modulefiles:

module load apps/intel-oneapi-2024.2.0/dl_meso/2.8_gpu
module load apps/intel-17.0/dl_meso/2.7.10
module load apps/intel-17.0/dl_meso/2.7.5
module load apps/intel-17.0/dl_meso/2.7.4
module load apps/intel-17.0/dl_meso/2.7
module load apps/intel-17.0/dl_meso/2.6

Running the application

You will notice that there are some differences between the User Manual and the CSF installation, in particular the naming of the executables. The tables below show the main executables that are available.
You can see full list of available executables, including utilities by running: ls -l $dlmeso_home/bin/ after loading the desired module.

DL_MESO v2.8 executables:

Executable Simulation
slbe.exe Serial LBE
plbe.exe Parallel LBE (uses MPI – single and multi-node jobs)
sdpd.exe Serial DPD
pdpd.exe Parallel DPD (uses MPI – single and multi-node jobs)
pdpd-omp.exe Parallel DPD (uses OpenMP – single-node multi-threaded jobs)
dpd_gpu_l40s.exe L40s GPU enabled version of DPD
dpd_gpu_a100.exe A100 GPU enabled version of DPD
dpd_gpu_h200.exe H200 GPU enabled version of DPD

DL_MESO v2.7 executables:

Executable Simulation
slbe.exe Serial LBE
plbe.exe Parallel LBE (uses MPI – single and multi-node jobs)
plbe-omp.exe Parallel LBE (uses OpenMP – single-node multi-threaded jobs)
sdpd.exe Serial DPD
pdpd-mpi-fftw-double.exe Parallel DPD (uses MPI & Double Precision FFTW3 – single and multi-node jobs)
pdpd.exe Parallel DPD (uses MPI – single and multi-node jobs)
pdpd-omp.exe Parallel DPD (uses OpenMP – single-node multi-threaded jobs)

DL_MESO v2.6 executables:

Executable Simulation
slbe.exe Serial LBE
plbe.exe Parallel LBE (uses MPI – single and multi-node jobs)
plbe-omp.exe Parallel LBE (uses OpenMP – single-node multi-threaded jobs)
sdpd.exe Serial DPD
pdpd.exe Parallel DPD (uses MPI – single and multi-node jobs)
pdpd-omp.exe Parallel DPD (uses OpenMP – single-node multi-threaded jobs)

Example Batch Jobs

Serial Batch job examples

Serial LBE batch job submission

  • Set up a directory from which your job will run, with all the required input files in it.
  • Write a job submission script, for example, in a file called jobscript:
#!/bin/bash --login
#SBATCH -p serial
#SBATCH -t 1-0

module purge
module load apps/intel-17.0/dl_meso/2.7

slbe.exe > OUTPUT
  • Submit: sbatch jobscript

Serial DPD batch job submission

  • Set up a directory from which your job will run, with all the required input files in it.
  • Write a job submission script, for example, in a file called jobscript:
#!/bin/bash --login
#SBATCH -p serial
#SBATCH -t 1-0

module purge
module load apps/intel-17.0/dl_meso/2.7

sdpd.exe > OUTPUT
  • Submit: sbatch jobscript

Parallel (multi-core) Batch job examples

It is highly recommended that you run scaling tests on 2,4,6,8,10,12,16,18,20,22,24,…40 cores before moving on to running larger jobs to see how well your job performs as the number of cores increases.

Parallel LBE batch job submission – 2 to 40 cores using MPI

  • Make sure you have the dl_meso and non-ib mpi modulefile loaded.
  • Set up a directory from which your job will run, with all the required input files in it.
  • Write a job submission script, for example, in a file called jobscript and asking for 8 cores:
#!/bin/bash --login
#SBATCH -p multicore_small
#SBATCH -n 8
#SBATCH -t 1-0

module purge
module load apps/intel-17.0/dl_meso/2.7

mpirun -n $SLURM_NTASKS plbe.exe > OUTPUT
  • Submit: sbatch jobscript

Parallel LBE batch job submission – 2 to 40 cores using OpenMP

  • Set up a directory from which your job will run, with all the required input files in it.
  • Write a job submission script, for example, in a file called jobscript and asking for 8 cores:
#!/bin/bash --login
#SBATCH -p multicore_small
#SBATCH -n 8
#SBATCH -t 1-0

module purge
module load apps/intel-17.0/dl_meso/2.7

export OMP_NUM_THREADS=$SLURM_NTASKS
plbe-omp.exe > OUTPUT
  • Submit: sbatch jobscript

Parallel DPD batch job submission – 2 to 40 cores using MPI

  • Set up a directory from which your job will run, with all the required input files in it.
  • Write a job submission script, for example, in a file called jobscript and asking for 12 cores:
#!/bin/bash --login
#SBATCH -p multicore_small
#SBATCH -n 12
#SBATCH -t 1-0

module purge
module load apps/intel-17.0/dl_meso/2.7

mpirun -n $SLURM_NTASKS pdpd.exe > OUTPUT
  • Submit: sbatch jobscript

Parallel DPD batch job submission – 2 to 40 cores using OpenMP

  • Set up a directory from which your job will run, with all the required input files in it.
  • Write a job submission script, for example, in a file called jobscript and asking for 12 cores:
#!/bin/bash --login
#SBATCH -p multicore_small
#SBATCH -n 12
#SBATCH -t 1-0

module purge
module load apps/intel-17.0/dl_meso/2.7

export OMP_NUM_THREADS=$SLURM_NTASKS
pdpd-omp.exe > OUTPUT
  • Submit: sbatch jobscript

DPD GPU batch job submission

  • The GPU enabled DPD executables are for running on single GPU only.
  • Only DPD can use GPU and this feature is only available in 2.8_gpu module
  • DEMO files are available in $APPDIR/DEMO directory.
  • Set up a directory from which your job will run, with all the required input files in it.
  • Write a job submission script, for example, in a file called jobscript and asking for 8 cores and 1 GPU:
#!/bin/bash --login
#SBATCH -p gpuL
#SBATCH -G 1
#SBATCH -t 1
#SBATCH -n 8

module purge
module load apps/intel-oneapi-2024.2.0/dl_meso/2.8_gpu

# Refer above 2.8_gpu module executable list, use dpd_gpu_XXXX.exe 
# version for the GPU type you have requested.
dpd_gpu_l40s.exe > OUTPUT

Last modified on October 6, 2026 at 6:46 pm by George Leaver