Skip to main content
SHARE
Blog

Titan has a very good year

Summit won’t be open to users for another three years, but let’s not forget that ORNL already has the world’s second-fastest computer—the 27 petaflop Titan.

Titan has been ranked either first or second in the world since researchers began using it in 2012. In fact, Titan delivered more computing power to researchers in the past year than ever before.

How fast is Titan? Imagine a million people doing one calculation each, every second, for a million years. When their time ran out, they would have done more than 30 million million million calculations. Titan would get through that workload in just under 20 minutes.

As a result, Titan has enabled breakthrough research from the practical to the profound. On the practical side, the largest, most accurate simulation to date of combustion will save both money and the environment by making our vehicle engines more efficient. Turning to the profound, simulations showing how our own Milky Way galaxy assembled over 10 billion years help explain how we got here.

Other researchers have broadened our knowledge of high-temperature superconductors, materials with the potential to rewrite the rules of energy production and distribution. Still others are expanding our knowledge in materials research, physics, biology and all areas of computational science.

They had these opportunities because the OLCF does a stellar job keeping Titan happy and productive. Not only were there no unscheduled outages last year—an impressive feat for a system of Titan’s size and complexity—but all 18,000-plus nodes stayed busy. On top of that, most of the simulations run on Titan—62 percent to be precise—were massive, taking up 20 percent or more of the machine.

“We’ve had a great year,” said OLCF Director of Science Jack Wells. “We’ve delivered more time—more capability—to users than ever before. We’ve delivered a higher percentage of big jobs than ever before, and at the same time we’ve had very high utilization. It’s hard to do a good job at both, but we’ve done that.

Related: