Webinar: Efficient (GROMACS+)CP2K compute resource usage for QM/MM simulation of biomolecular systems (2022-05-24)



Hybrid quantum/classical (QM/MM) simulation of biomolecular systems has grown in popularity in recent years, facilitated in part by the availability of software such as the GROMACS-CP2K and GROMACS-CPMD interfaces developed within BioExcel, which couple GROMACS with the CP2K and CPMD quantum chemistry / electronic structure codes.

Performing productive QM/MM simulation of biomolecular systems requires researchers to not only carefully choose suitable QM and QM-MM coupling treatments – a topic covered in depth in our workshop on Best Practices in QM/MM simulation of Biomolecular Systems – but also to consider how best to make use of available computational resources, which are needed mainly for costly QM calculations. Available computational resources may constrain the choice of treatment that is feasible and hence the quality of results obtained, which may amount not merely to a quantitative difference in accuracy but even to qualitatively different conclusions regarding structure or function.

Understanding how the QM/MM simulation software employed can best make use of available hardware including processors (CPUs) and accelerators (GPUs) has the potential to not only improve efficient utilisation of these resources but also to significantly accelerate research and individual researcher productivity. Moreover, knowing what computational resources would be needed to pursue a given treatment helps justify and obtain access to high-performance computing (HPC) resources at an institutional, regional, or national level, including the largest facilities available to researchers in the form of – in the EU – supercomputing infrastructure provided through PRACE and EuroHPC.

With that in mind and in order to assist researchers using or considering using CP2K – with GROMACS or as a standalone application – for QM/MM simulation of biomolecular systems, we provide in this webinar a brief overview of the parallel performance of CP2K on hardware commonly found on modern HPC machines for a range of biomolecular systems and QM treatments that together make up the BioExcel QM/MM Benchmark Suite. This overview is intended to function as a rough reference and a useful starting point to guide researchers towards obtaining good performance on available hardware for their system of interest, and/or to help plan future work that includes access to and increased usage of HPC resources.


Holly Judge

Holly is an applications consultant at EPCC, the University of Edinburgh. She has a research background in computational chemistry and physics, but now in her role at EPCC she is mainly focused on exploring the performance of high performance computing (HPC) applications used in  scientific research. In particular, within her work for BioExcel she has been looking at the improving the performance of QM/MM calculations in CP2K. She also is part of the CSE team for the UK national supercomputer service – ARCHER2.

Arno Proeme

Arno is a software architect / research software engineer at EPCC at the University of Edinburgh and has a background in computational statistical physics research. In his role at EPCC he has worked on a number of projects improving the parallel performance of research software to take advantage of modern computing architectures, as well as being involved in training and supporting researchers using HPC facilities such as the UK’s national supercomputing service. He also teaches on EPCC’s MSc in HPC. Within BioExcel he coordinates EPCC’s contributions including software development / performance optimisation work, as well as BioExcel’s overall community support and engagement activities.

Register for webinar

Title: Efficient (GROMACS+)CP2K compute resource usage for QM/MM simulation of biomolecular systems

Date: May 24, 2022
Time: 15:00 CEST

Register URL: https://us02web.zoom.us/webinar/register/WN_OuECCE8eSbuPd5UMZEgOWw

After registering you will receive an email with details of how to connect to the webinar.