turbomole

Description

TURBOMOLE is an ab initio computational chemistry program that implements various quantum chemistry methods.

Restricted use

Apply for access: https://doku.pc2.uni-paderborn.de/pages/1902360/Licensed+Software

More information

- https://www.turbomole.org/ - Refer to $TURBODIR/DOC after loading the module - Call TmoleX to start the GUI

Available Versions of turbomole

Version	Module	Available on

Version	Module	Available on
tmolex24	chem/turbomole/tmolex24	Noctua 1, Noctua 2
tmolex23	chem/turbomole/tmolex23	Noctua 1, Noctua 2
tmolex22	chem/turbomole/tmolex22	Noctua 1, Noctua 2
tmolex16	chem/turbomole/tmolex16	Noctua 1, Noctua 2
7.8.1	chem/turbomole/7.8.1	Noctua 1, Noctua 2
7.8	chem/turbomole/7.8	Noctua 1, Noctua 2
7.7.1	chem/turbomole/7.7.1	Noctua 1, Noctua 2
7.7	chem/turbomole/7.7	Noctua 1, Noctua 2
7.6	chem/turbomole/7.6	Noctua 1, Noctua 2
7.1	chem/turbomole/7.1	Noctua 1, Noctua 1, Noctua 2, Noctua 2

This table is generated automatically. If you need other versions please click pc2-support@uni-paderborn.de.

Usage Hints for turbomole

If you need support in using this software or example job scripts please click pc2-support@uni-paderborn.de.

How to start TmoleX

Proceed as for starting GaussView but load a TmoleX module instead of a Gaussian module. The TmoleX command then starts the GUI, e.g.:

$ ml chem/turbomole/tmolex24 
$ TmoleX

Then you can create project with new molecules by clicking “Import Coordinate File”, “Open 3D Molecular Builder”, “Draw 2D”, or “SMILES to 3D” as shown in the figure below. More detailed usage can be found in $TURBODIR/DOC after loading a turbomole module, e.g. ml chem/turbomole/7.8 or the TmoleX documentation at TURBOMOLE.

How to create Turbomole input

For simple calculations it is convenient to create the input file control by using a text editor, e.g. vim.

$atoms
   basis = def2-TZVP
$coord file=coord
$dft
   functional pbe0
$rij
$title Geometry optimization (RIJ-PBE0 with def2-TZVP)
$end

In addition, Turbomole provides the define command, which is an interactive input generator, for more complex input file. Also you can use the TmoleX GUI program (see above) to create the Turbomole input file.

How to run Turbomole calculations

Some of the Turbomole modules are parallelized by MPI and/or OpenMP. Example Slurm jobscripts for OpenMP and MPI parallelization on a single node are shown below. The MPI parallelized modules are not recommended for performance reasons (see 3.4 Parallel Runs in the Turbomole User’s Manual, Version 7.8). This is also confirmed by our tests on Noctua 2.

run OpenMP parallel jobs

For the created Turbomole input and coordinate files, the Slurm jobscript below runs an OpenMP parallelized RI-DFT calculation on one compute node of Noctua 2 using 128 CPU cores.

Note: We recommend the execution of Turbomole using the OpenMP parallelization on one CPU node.

#!/usr/bin/env bash
#SBATCH --job-name=tm_omp
#SBATCH --time=00:30:00
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=128
#
module reset
module load chem/turbomole/7.8
#
# parallel SMP environment
#
ulimit -s unlimited
export PARA_ARCH=SMP
export PARNODES=${SLURM_CPUS_PER_TASK}
export TM_PAR_OMP=on # use OpenMP parallelized modules
export OMP_PLACES=cores
export OMP_PROC_BIND=close
#
# perform the RI-DFT calculation
#
workdir=/dev/shm/tm_omp_${SLURM_JOB_ID}
mkdir -p ${workdir}
cp control coord ${workdir}; cd ${workdir}
jobex -ri -np ${SLURM_CPUS_PER_TASK} -outfile ${SLURM_SUBMIT_DIR}/jobex.out
cp ${workdir}/* ${SLURM_SUBMIT_DIR}

run MPI parallel jobs

For the created Turbomole input and coordinate files, the Slurm jobscript below runs an MPI parallelized RI-DFT calculation on one compute node of Noctua 2 using 128 CPU cores.

run GPU enabled jobs

For the created Turbomole input and coordinate files, the Slurm jobscript below runs a GPU enabled RI-DFT calculation on one GPU node of Noctua 2 using 128 CPU cores and 1 NVIDIA A100 GPU.

The following output in jobex.out confirms the usage of the A100 GPU in the calculation.

Please note:

known limitation: RI-J integrals are done on the CPU, so using many cores on CPUs can be more efficient than to use CPU and GPU but with less CPU cores.
more detailed information for GPU enabled Turbomole can be found in ${TURBODIR}/DOC/README-GPU.txt.

Further information

Turbomole documentation: https://www.turbomole.org/turbomole/turbomole-documentation/
Turbomole users forum: https://forum.turbomole.org/