To protect against bit rot, I'll quote the whole post [1]:
First light on Medusa Jan 24 2009, 10:42
Medusa from the back - what a mess!
I've just completed my first successful set of calculations on
Medusa, the new computer cluster that the Theory Group recently
purchased. Medusa comprises 72 nodes of 8 AMD Opteron processors
each, linked together with an InfiniBand networking fabric. There
have been a few teething problems, but everything seems to be
working OK now.
The test calculations involve Monte-Carlo simulation of radiative
transfer through the electron-scattering magnetosphere of σ Ori
E. I'm trying to reproduce the temporal variations in the star's
linear polarization, originally observed by Kemp & Herman
(1977). I've made some Stokes IQU images of the simulation results,
and strung them together as a movie (AVI; MOV; movie help).
The image is linked from the post [2].
[1]: http://www.astro.wisc.edu/~townsend/news.php#medusa-first
[2]: http://www.astro.wisc.edu/~townsend/resource/news/medusa-cables.jpg