Funded by


Subproject D.6

Accelerating MD using GPUs

Wie können Simulationen effektiver berechnet werden?

The objective of this project is to develop simulation algorithms for the systematic investigation of germ formation - the early state of crystal growth - in systems with charged particles such as proteins.

To substantially speed up Molecular Dynamics simulations of systems with large numbers of particles, we devise novel algorithms that are optimally suited for Graphic Processing Units (GPUs). Our main target are computationally intensive interactions, namely hydrodynamics and electrostatics, which are by far the most expensive algorithms. These algorithms we will extend to scale efficiently on massively parallel architectures (for example, GPUs).

Further information about this subproject


A turbulent flow developing behind an obstacle. Simulation was using a GPU-based lattice Boltzmann solver.