The HEC hosts serial and SMP parallel versions of the Turbomole ab initio Electronic Structure Calculations package. The package is licensed to Andrew Kerridge in the Chemistry Department.

Turbomole License availability

The Turbomole licence is held for the Chemistry Department by Andrew Kerridge. Please contact him to authorise access to the software.



More information

 View available versions of Turbomole

To view the currently available Turbomole versions, enter the following command:

module avail turbomole


You will see output similar to this:

----------- /usr/shared_apps/Modules/apps ------------
turbomole/6.6(default) turbomole/6.6-smp


The default is the serial version.




 A sample batch job script for serial jobex

Serial turbomole batch jobs may be run by creating a batch job control script (for instance, called jobex.com) like the following:

#$ -S /bin/bash

#$ -q serial
#$ -l h_vmem=1G

source /etc/profile
module add turbomole/6.6

jobex


The final line in the script is a call to the relevant turbomole script (in this case jobex). Note that no additional environment variables need to be set up - this is all handled by the turbomole module file. The script is then submitted for running by typing the following from the directoiry containing you job's data files:

qsub jobex.com


The job submission template above follows the same template as a general serial batch job.




 A sample batch job script for SMP parallel jobex

For job types which support SMP parallel versions of Turbomole applications, the following template job script can be used:

#$ -S /bin/bash

#$ -q parallel
#$ -l node_type=10Geth*
#$ -l np=8
#$ -l h_vmem=1G

source /etc/profile
module add turbomole/6.6-smp

jobex


Job submission of this script is identical to the serial version. Note that no extra environment variables pertaining to parallel running need to be set up - the module command will inherit and set up correct values automatically from the parallel job environment.




 A sample batch job script for serial NumForce

The template for serial jobex (above) can be used, replacing the call to jobex with an appropriate call to NumForce.


 A sample batch job script for SMP parallel NumForce

The recommended approach for parallelism for NumForce is slightly different. The following script serves as a template:

#$ -S /bin/bash

#$ -N methyl-0
#$ -q parallel
#$ -l node_type=10Geth*
#$ -l np=8
#$ -l h_vmem=2G

source /etc/profile
module add turbomole/6.6-smp

TM_MFILE=tm_mfile.$JOB_ID.$SGE_TASK_ID
parhosts > $TM_MFILE

NumForce -central -ri -mfile $TM_MFILE > NumForce.out


The additional lines create an mfile which tells NumForce how many processors/cores it should run on.




 SMP Turbomole Guidelines

Not all Turbomole tools are multi-core, and not all multi-core tools may be efficient with large numbers of cores - please refer to the Turbomole guide for details. To prevent wasting resource on the HEC, SMP Turbomole should only be used for workloads that can efficiently use the number of requested cores.

The template above is intended as an example. The template can be modified in the same way as an OpenMP parallel job. Please refer to that page for an explanation of the job directives contained in the above template.




Requests

For access to the Turbomole license:
Contact Andrew Kerridge

Further advice

Documentation can be found on the Turbomole Website.