The CRA has posted a number of interesting videos related to computer science research and how it has changed the world. Take a look at

http://www.youtube.com/user/computingresearch

to see the videos. The "Human Computation" session was one I particularly enjoyed.

Continuing with the HPC/HTC (high throughput computing) theme, the MSCS department deployed their first "large scale" Condor pool over break. We now have over 100 slots available for distributed computations on lab machines throughout the department.

For our first test, we tried running a simple shell script. This failed, until I discovered it was a text file in DOS format. This caused the Linux execution host to try to execute

/bin/bash\r

instead of

/bin/bash

and Condor failed to execute with a file not found error.

It's amazing how even after almost 30 years of computing, I run into the same old problems.

Over the past 2 years, my interest in computational clusters, grids, and cloud computing has grown. First this interest grew out of my own need for more computational power for my text mining research. My interest was sustained because the need for research computing in Southeast Wisconsin.

Last year, I got word of the Microwulf project directed by Joel Adams. This project aims build machines with very low dollar/FLOP and watt/FLOP ratios. I saw this as an opportunity to create some computational power, but also as a tool to get students interested in the more complex systems they will face in the future.

With the help of Adam Koehler, Michael Schulz, several members of MULUG, Marquette's version of the Microwulf, Slayer now exists. It sports 55.76 GFLOPS for just under $2000. We haven't yet tapped the 4 NVIDIA GPUs yet, which should add many more FLOPS.

We'll be turning our focus to utilizing Slayer to get some real work done as well as some fun projects. Slayer will also be making its way to local schools to recruit students into computer science and other technology related majors.