These days I work mostly on building planets from dust using a 2-phase parallel dusty SPH code that I wrote. Co-planeteers are Sarah Maddison and James Murray, both at Swinburne which is where I am about 1/2 the time.

N-body and Smoothed Particle Hydrodynamics asynchronous MPI parallel codes on as many processors as are available. Attempts at good load balancing. Caching (well, more re-use really) of SPH neighbour lists, caching of parallel tree data, caches of caches... Optimal choices of h and minimal comms ways of altering it. Speed, speed, speed. Lists of lists of lists...

All this is made possible by the stylish algorithms of Salmon and Warren who's codes (sorry, codez) have (pretty much) yet to be surpassed in a decade of parallel astro computations. Dey are da godz (kinda).

Another of da godz (and who is thankfully still in the field and pushing back those frontiers) is John Dubinski who I'm (sometimes) working with on new mega codez Tree/PM + SPH.

Cell-cell forces, heterogeneous dynamic load balancing, and multi-timestepping in development. Occasional building of Beowulf clusters. Some OpenGL. Linux/Alpha is cool. Linux in general is cool (apart from no NFSv3 (update, 2001: works), NIS bugs, basically x86 only compilers, no page colouring (update 2003: 1/2 there), missing man pages... blah blah blah). RAID and journalling filesystems are gladly welcomed to the fold. I'm a huge old SGI fan too but don't tell anyone. Far too much computer and sysadmin knowledge for my own good. Far too few publications.

Vodka helps. RC51 would help too, but I can't afford it and the rest are too slow and don't make enough noise. Well, except for my ace VFR750 that has taken much abuse and blatant hoonage in its stride. Gotta love a sunburnt country (except for the melanoma). I aspire to better pretty pictures.

And yes, if anyone asks, kangaroos hop down the main street.