Pushing the Envelope
Four Out of Six Gordon Bell Finalists Used Titan
Four out of six Association for Computing Machinery’s Gordon Bell Prize finalists used Titan to overcome complex computational challenges in their fields. From the dynamics of millions of crowded proteins to the billions of particles that radiate from passing plasma jets, ultra-high-resolution simulations on Titan make for strong contenders. The Gordon Bell Prize is awarded for achievement in applying high-performance computing to applications in science, engineering and large-scale data analysis.
“Taking a Quantum Leap in Time to Solution for Simulations of High-Tc Superconductors,” led by Peter Staar of ETH Zürich
For use on Titan’s hybrid CPU-GPU architecture, researchers developed a new quantum cluster algorithm DCA+, an extension of the Dynamical Cluster Approximation, to simulate models of high-temperature superconductors. Because conventional superconductors require near-absolute zero-temperatures to operate, which is expensive and difficult to maintain, researchers want to understand the complex physics behind what makes high-temperature superconductors lose their electrical resistance.
“20 Petaflops/sec. Simulation of Proteins Suspensions in Crowding Conditions,” led by Massimo Bernaschi of ICNR-IAC Rome
An application for studying the behavior of proteins in crowded conditions reached 26 petaflops on Titan during the most intense calculations. With a sustained performance of 20 petaflops, the team simulated the structure and dynamics of proteins in biofluids—an elaborate configuration of millions of proteins moving on multiple time scales under different conditions. Ultimately, these simulations contribute to a better understanding of cellular function.
“Radiative Signatures of the Relativistic Kelvin–Helmholtz Instability,” led by Michael Bussmann of HZDR-Dresden
The Kelvin–Helmholtz Instability (KHI) is turbulence that occurs when two streams of plasma flow past each other. Understanding the dynamics of the KHI has implications for astrophysics, fusion, and laser-plasma physics. By mapping radiation emitted by electrons during the KHI, researchers can analyze the plasma dynamics taking place, something difficult, or, at times, impossible to observe directly. Titan was able to simulate the 75 billion protons and electrons that make up this radiation signature.
“HAAC: Extreme Scaling and Performance across Diverse Architectures,” led by Salman Habib of Argonne National Laboratory
Researchers developed the Hardware/Hybrid Accelerated Cosmology Code (HAAC) to achieve high-scalability and sustained performance on petaflop machines such as Titan and those at other Department of Energy laboratories. In order to simulate cosmological surveys, which must incorporate physics models as well as mounting observational data, researchers will need machines that perform on the order of petaflops or beyond to achieve the accuracy needed for further study. HAAC was able to evolve 1.1 trillion particles on Titan.
The Gordon Bell Prize is an award presented by the Association for Computing Machinery (ACM) each year in conjunction with the SC Conference series (formerly known as the Supercomputing Conference). The prize recognizes outstanding achievement in high-performance computing applications. The Gordon Bell Prize will be announced at SC13 held from November 17-22 in Denver, CO.
—Katie Elyce Jones