Moore's Law says that the speed of computers will double every 18 months. Even if that trend continues, the demand for computational speed in some areas of science and industry may grow even faster. To meet this demand, scientists are starting to utilize clusters for their computing more than ever before.

“Fifty-eight percent of the computers in recent Top 500 Supercomputer Sites are clusters. This is up from only thirty percent a year ago." says Tom Murphy, computer science faculty member at Contra Costa College (CCC) and director of the college’s High Performance Computing Regional Education and Training Center. "With the proliferation of clusters, the need for technicians who can provide support will continue to grow in the near future."

This is the starting point for the new high performance computing program at CCC. Berkeley Lab has partnered with CCC in preparing students for entry-level jobs as technicians who will administer Linux cluster support.

"This is a really unique and exciting program that addresses a growing field," said Charlie Verboom, desktop support manager for the Information Technologies and Services Division (ITSD) and the team's sponsoring mentor. “It’s a field that offers immediate job growth.”

The program is being brought closer to home with a summer internship program that gives students an opportunity to get hands-on experience running clusters at Berkeley Lab. ITSD is hosting a faculty and student team from CCC. As part of the Lab's Center for Science and Engineering Education (CSEE), the Faculty and Student Teams Program is a cooperative effort between the Department of Energy (DOE) Office of Science and the National Science Foundation (NSF).

"The internship assists students with furthering their education goals,” Charlie said. “And most of them go on to four-year schools.”

Tom is working hand in hand with the students at the Lab as their faculty mentor. He wants to educate students that clusters are having an impact on everyday life: Pixar used clusters to make “Finding Nemo,” and AskJeeves uses clusters to handle Web queries.

Along with Berkeley Lab, Pixar and biotech companies are among CCC's high performance computing program’s advisers. But Berkeley Lab is uniquely involved in partnering with the program by assisting with the development of the internship program. In an effort to fill the future job market with qualified technicians, the program trains students in system and network administration, security and parallel programming.

CCC is in the process of getting state certification to be the first community college in California to offer an Associate of Science degree in high performance computer technology. The high performance computing program begins this fall, and requires students to do an internship, as the students are doing this summer. It’s a two-year program with no prerequisites, although a passion for math and computer hardware and software is strongly recommended. Tom is working hard to extend this type of program to the 26 other Bay Area community colleges, thus allowing students to transfer credits. He has assembled a 64-node cluster and a 16-node cluster at CCC and made them available to other math and science courses.

Before the students could begin their individual work on clusters, they needed to become familiar with Linux. So they began their summer internships with a crash course in Linux using ITSD's Wale Soyinka's lab manuals. (See story on Wale’s books also in this issue.) Now all of the students are developing lab exercises, with the help of Tom and Wale. These exercises will be used in the introductory high performance computing class that Tom will be teaching next month.

For profiles of the four students, please see below:

Bassam first became interested in high performance computing when he attended a conference earlier this year and saw a simulation of a particle problem. He decided he wanted to do a similar problem and match his skills against those of an expert.

"Anything that comes towards me, I just do it," Bassam said. "I'll use these skills in my career in the future. I want to learn how to use clusters as a tool to work for me."

So he joined Tom's group and is spending the summer at Berkeley Lab parallelizing code and running it on a cluster. Bassam is also helping to develop lab exercises on high performance computing that will be used in future classes at Contra Costa College.

Bassam attended Contra Costa College for two years and is transferring to UC Davis this fall to major in civil engineering. He hopes to pursue a career in water resources and development.

Adam has a strong math background and became interested in computers when he took a C++ class taught by Tom. Programming an algorithm helped him understand math better and he's been interested in seeing what computers can do ever since.

"I'm not a big fan of computers, but I'm a big fan of the results," said Adam, who is also an artist and welder. "For me computers are a means to an end."

This summer Adam is working with the Linpack benchmark, which is used to determine the performance of the world's fastest supercomputers that make the Top 500 list. He's running the benchmark on the student-built clusters and is learning what it takes to achieve a high gigaflop/s rating.

After three years at CCC, Adam is taking a semester off and deciding whether or not to apply to UC Berkeley. Eventually, he wants to be a community college math teacher. "I want to give something back," he said. "I think community colleges give an education to a lot of people who otherwise wouldn't have the opportunity."

Martin, who has attended CCC for one year, plans to work in the field of nuclear engineering and will attend UC Berkeley this fall to study just that.

"High performance computing is used in nuclear engineering, and I wanted to find some piece that would allow me to build a parallel application," said Martin, who served as a nuclear operator in the Navy for six years before returning to school.

When he learned that UC Berkeley's nuclear engineering program is using a cluster to model reactor physics, he decided that he wanted to run a simulation of his own. This summer at the Lab he's running a simple Monte Carlo simulation. He got the idea from Peregrine, a 3D Monte Carlo simulation that is used in calculating doses for radiation treatment for cancer patients.

"Because you can't put a detector in someone's brain to determine the necessary beam intensity, for example, you have to use a simulation," Martin explained. "I'm modeling photons, which don't interact with each other so they're particularly well-suited to run on a parallel cluster."

Ben enjoys figuring out why computers crash, and how to resuscitate them. He loves the challenge of getting around software and hardware problems and fixing them.

"I've been using computers all of my life--since I was three," said Ben, who will be a senior this fall at Hayward High School.

As Charlie Verboom's high school summer intern, Ben wanted to learn Linux and set up clusters. Ben's interests merged with the faculty/student program, so he joined the group. Now he's setting up one of the clusters that the other students are using. He hooked up a number of unused desktop computers and got them running again as a cluster using the Berkeley Lab developed Warewulf Cluster toolkit. Ben is also working on setting up a cluster model that can be used for students at Contra Costa College next fall when they set up a cluster.

Ben is deciding between pursuing a career in the computer field, nanotechnology or physics.