http://loki-www.lanl.gov/
http://cesdis.gsfc.nasa.gov/linux-web/beowulf/beowulf.html
and some fun software packages to run on them
http://www.ai.uga.edu/students/jae/ai.html
http://sal.kachinatech.com/ # try searching for molecular nanotech
http://world.std.com/~wware/ncad.html
Actually, that last earl there is the most interesting one. Here's a
chunk of text from the page:
>Here's My Bias about Nanotechnology Development
>
>I think the safest way to develop nanotechnology is software-first,
>hardware-last. Here is why I think that: We can't actually do real
>damage until the hardware is in place. By having a lengthy, detailed,
>and well-informed public discourse on what would be possible, prior
>to the existence of any hardware, we can make intelligent policy
>decisions. These decisions must precede the existence of potentially
>dangerous hardware. (By hardware, I mean whatever physical mechanisms
>allow us to manipulate atoms and molecules with sub-nanometer
>precision, and of which a self-replicating Drexlerian assembler
>would be the riskiest example.)
>
>One of the most frightening scenarios is that nanotechnology is
>developed by a small group, which refuses to share it with the rest
>of the world. The world is already polarized into "haves" and
>"have-nots", and this polarization is reflected in political tensions
>all over the globe. Policy discussion must be inclusive, not
>exclusive. The tools to participate in that discussion (including
>simulators, if they prove valuable) should be available to everybody,
>because if nanotechnology ever really arrives, everybody will be
>effected.
end quote
-- r q watt