Navigation and service

TetraMAG - Our micromagnetic code

Finite element micromagnetic simulations are a powerful tool for the study of magnetization dynamics in samples with arbitrary geometries. However, resolving the dynamics in structures that span a wide range of length scales (from nanometers to several microns) can require large amounts of computing resources and long computation times. This is why we have ported our TetraMAG micromagnetic simulator to run on Graphical Processing Units (GPUs) using the CUDA software platform. By efficiently exploiting the massively parallel GPU architecture, a significant speedup of the micromagnetic simulations is achieved with double precision accuracy, compared to parallel computations on CPUs. This development allows to conveniently perform systematic large-scale and high-resolution micromagnetic studies such as for example of domain wall motion in long nanowires or of cross-tie domain wall dynamics in mesoscopic structures. Our GPU-based finite-element code enables high performance computing on a personal workstation.

A. Kákay, E. Westphal and R. Hertel, IEEE Trans. Magn. 46, 2303-2306, 2010