Four Science “Titans” Riding Our GPUs to Scientific Glory

by Brian Caulfield

There’s hard core, and then there’s hard core.

Battling through “Assassin’s Creed III” on a PC with a pair of GeForce GTX Titan GPUs is one kind of hard core.

Simulating the behavior of superheated gasses ejected from stars into the cosmos on a supercomputer equipped with more than 10,000 GPUs is quite another.

This is the kind of research that moves our civilization forward. And the scientists pushing thousands of our GPU accelerators to their absolute limit in pursuit of this knowledge are – plain and simple – rock stars.

They’re the reason why the Gordon Bell Prize, considered the “Nobel Prize of supercomputing,” was created. Awarded every year at the annual Supercomputing Conference (SC), this award recognizes outstanding scientific achievement on the biggest and most powerful supercomputers.

The next Gordon Bell winner will be crowned at SC13, which will be held mid-November in Denver, Colorado.

And four of the six finalists conducted their research on the Titan supercomputer, which is run by the Oak Ridge Leadership Computing Facility at Oak Ridge National Laboratory, the nation’s number-one scientific supercomputer, open for use by researchers around the world.

A monster system equipped with 18,688 of our Tesla GPU accelerators, Titan has been powering breakthroughs across a broad range of scientific domains in the year since it was announced at SC12.

Given that there are only six finalists vying for this prestigious honor, we like the odds that Titan and GPU accelerators might be part of the winning entry.

All four of the Gordon Bell researchers will be presenting will be speaking at the GPU Technology Theater in our SC13 booth at the conference (Booth # 613), giving a sneak-preview of their remarkable work.

Here’s a preview of what you can expect:

Thinking Small
Tuesday, Nov. 19 — 5:30 – 6:00 PM Mountain Time

Simone Melchionna, a researcher at the Institute for Chemical Physical Processes at the National Research Council of Italy, will review supercomputer-generated simulations of the behavior of proteins – the engines of all life – inside the crowded conditions of living cells. Melchionna and his colleagues mixed new mathematical models, computational algorithms, and more than 18,000 GPU accelerators to simulate the behavior of roughly 18,000 proteins for the first time.

What a Gas
Wednesday, Nov. 20 — 12:30 – 1:00 PM Mountain Time

Michael Bussmann, from Helmholtz-Zentrum Dresden-Rossen, a German research laboratory, will discuss his findings on what happens when streams of hot gas are ejected from stars, black holes, and galaxies. Using Titan, Bussman simulated the particle dynamics and radiation emitted during the formation of Kelvin-Helmholtz instabilities in the particle flows generated by those streams.

Simulating Superconducting
Wednesday, Nov. 20 — 4:00 – 4:30 PM Mountain Time

Understanding the behavior of super-conductors has been a major challenge to physicists for a generation. Peter Staar, a PhD student, at Eidgenössische Technische Hochschule Zürich, is using enhanced algorithms – and a new generation of  supercomputers able to crank through vast numbers of calculations efficiently– to crack one of the most sought-after problems in high-temperature superconductivity: at what temperature a material transitions into a superconductor.

Thinking Big
Thursday, Nov. 21 — 1:00 – 1:30 PM Mountain Time

Salman Habib, a senior scientist at the Argonne National Laboratory, threw GPUs at the task of simulating the evolution of the structure of the universe. He’ll talk about how he used code tuned to run on machines equipped with both GPUs and CPUs – like Titan – to dramatically increase the performance of these sprawling simulations.



These are just a small sampling of the GPU-accelerated science presentations we will be hosting in our GPU Technology Theater.

From GPU-accelerated climate modeling to cosmetics to research into flying snakes, our theater will be booked solid with an amazing lineup of industry luminaries, scientists, and developers talking about all the ways GPU acceleration is driving their research.

You can view the full SC13 GPU Technology Theater schedule here.

Not going to SC13 this year? No problem. All of our booth presentations will be streamed live at our SC13 web page.