Firms Face Geek Gap

OAK RIDGE, Tenn. -- The private sector is struggling to bridge a supercomputing skills gap, according to attendees at the High Performance Computing User Forum taking place this week at Oak Ridge National Lab.

The engineers and scientists who are talented in the field of supercomputing are more interested in making a mark than earning big bucks. Hence, enterprises are losing the best talent to academia and research, prompting concern from execs at the Tennessee event.

It makes it harder to compete,” admits a high performance specialist from a government contracting firm, who asked not to be named. “If you’re not fast enough, you don’t get the business.”

The skills shortfall can pose other problems, according to the exec. “If you lose key people and you can’t rapidly re-staff, you could run into [government contract] penalties for delays.”

Competition for the best supercomputing brains is particularly intense in the United States, which is home to a myriad of big-name research labs. As well as Oak Ridge and Los Alamos National Lab, other high profile sites include Sandia National Labs and the Lawrence Livermore National Lab.To make matters worse, a number of American universities are developing their own cutting-edge high performance computing projects (see Pittsburgh Picks Big Ben, AMD Powers Rice , and Univ. of Texas Teams Up With Sun).

Opportunities at government and university research centers attract the kind of talent that favors science over corporate ladder-climbing. “Certainly, you can make a lot more money in the private sector, but there’s probably less opportunity for core science and innovation,” says David Torgersen, associate director for platform computing at Pfizer Inc. “A lot of people feel that they can work with broader applications in academia."

To attract geeks, Pfizer and other firms are forging new partnerships with U.S. government labs and universities.

Other organizations are looking at their fundamental approach to supercomputing. “If companies want to overcome the supercomputing skills gap, they need to look at how supercomputing is perceived within their own business,” says Andrew Jones, high end computing development manager at Manchester University in the U.K, where the world's first stored computer program was developed in 1948. He went on to say that if supercomputing is not seen as underpinning the company’s business, then it will be hard to attract the best people.

— James Rogers, Site Editor, Next-Gen Data Center Forum0