Efficient use of Molecular Dynamics simulation applications in an HPC Environment@CINECA
→
CET
CINECA - BOLOGNA
CINECA - BOLOGNA
via Magnanelli 6/3
40033 Casalecchio di Reno (BO)
Italy
Description
Classical molecular dynamics simulation programs are very widely used on supercomputers. Nowadays, computational chemists rarely write their own codes but instead exploit the opportunities offered by high quality, freely available and open source packages such as Gromacs, NAMD, LAMMPS or DL_POLY. But although researchers are usually able to sucessfully construct the necessary inputs to model their systems accurately, actually using the programs in an efficient way on current HPC facilities can be more challenging. Part of the reason lies in the fact that researchers rarely study the algorithms used to optimise and parallelise the programs and so do not choose the optimum set of program options and inputs for a given HPC computer. And of course, the complex nature of current HPC systems, many of which are now equipped with accelerators such as GPUs or MIC cards, further adds to the number of parameters which researchers must consider before launching their simulations.
In this course we describe and illustrate with practical examples the common optimization and parallelization techniques implemented in popular molecular dynamics programs such as Gromacs and NAMD. The aim is to give researchers the tools to run classical molecular dynamics simulations in the most efficient way possible on current and future HPC systems.
Target Audience
Research interest involving classical molecular dynamics in computational biology, chemistry or biophysics
Topics
• Basic principles of classical Molecular Dynamics (MD).
• Parallelisation of molecular dynamics with particle, force and domain decomposition methods and the Particle Mesh Ewald algorithm.
• Benchmarking and performance analysis of MD codes.
• Factors limiting parallel scaling.
• Advanced optimisation and parallelization features (e.g. GPUs) available for programs such as Gromacs and NAMD and their use on modern HPC systems.
• Project design and strategies for resource application.
About half of the course will consist of practical, hands-on sessions. The programme will include invited talks from two guest speakers working in the field.
Benefits
By the end of this course students should be able to:
• understand the algorithms driving modern classical molecular dynamics programs;
• run common programs such as NAMD or GROMACS efficiently on today’s HPC systems;
• improve their prospects when submitting project applications for requesting resources from providers such as PRACE or other agencies.
Grant:
A grant of 300 EUR (for foreign students) and 150 EUR (for Italian students) will be available for participants not funded by their institution and not working in the Bologna area.
Some documentation will be required and the grant will be paid only after a certified presence of minimum 80% of the lessons and about 1 month after the ending of the course.
For further information about how to submit for the grant, please wait the confirmation email that you are accepted to the course about 3 weeks before the date of the beginning of the lessons.
The lunch for the 3 days will be provided by Cineca.
Course Organizers:
Andrew Emerson (Cineca)
Giovanni Chillemi (Cineca)
Alessandro Grottesi (Cineca)
ANNOUNCE: WE REACHED THE MAXIMUM NUMBER OF SUBSCRIBERS. PLEASE WRITE AT SUPPORT (corsi@cineca.it) TO BE ADDED TO THE RESERVE LIST. WE WILL CONTACT YOU IN CASE OF SEATS AVAILABLE.