How to Speed Up OOMMF

By | February 24, 2015

OOMMF can be incredibly useful for studying micromagnetic systems. Some calculations, however, can take an incredibly long time.

Check out these tips for how to speed up OOMMF calculations.

1. Use more (and faster) cores

This is an easy one. If you need you computation to go faster, use a faster computer. If you have access to HPC facilities, you can probably install OOMMF and use it there.

See my previous post on how much speed increase you can expect by using more cores in OOMMF.

2. Increase the cell size

Another easy way to decrease calculation time is to increase the cell size of the simulation. You have to be careful though, as if you increase the cell size too much, you could go beyond the exchange length, which can introduce errors.

See more on simulation cell size here.

3. Make your calculation dimensions a power of 2

OOMMF uses fast fourier methods for some of its calculations, which means you can speed up calculations by setting the x, y and z dimensions of the simultion region to be a multiple of 2.

4. Try out different Evolvers

There are two main evolvers in OOMMF: Euler and RungeKutta. In many situations the RungeKutta method will be faster, so be prepared to use that instead of Euler. Be careful though, as there can be some situations where Euler comes out on top.

5. Use periodic boundary conditions

Have a think about the problem you are trying to simulate. Is there anyway you can divide up a larger problem into a periodic mesh? This effectively reduces the volume that you actually need to simulate, which can dramatically decrease simulation times.

Check out my post on periodic boundary conditions in OOMMF.