By Jessen O'Brien
Photo courtesy of Argonne National Laboratory
“ Mira’s petascale computing capabilities are like hitting the accelerator on the time-to-solution for a larger and more complex class of investigations.”
Division director of the Argonne Leadership Computing Facility
Mira, the fifth-fastest supercomputer on Earth, can simulate everything from the earliest evolution of the universe to the future effects of climate change. It also can compress time by churning out massive calculations in a tiny fraction of the time it would take an average laptop to do them.
“Mira has 200,000 times more computing cores than a standard laptop, so what would take one hour to do on Mira would take 20 years on a laptop, assuming the problem could fit in the laptop’s memory,” says Paul Messina, Director of Science at the Argonne Leadership Computing Facility.
Mira was officially dedicated on July 1, 2013, in a ceremony that included U.S. Sen. Dick Durbin and Argonne Director Eric D. Isaacs. Durbin described the Argonne supercomputer as “one of Illinois’ and the country’ great assets.”
Weighing about 100 tons and with the raw computing power of 58 million iPads, Mira might sound unfathomable. However, history has shown that today’s supercomputers can pave the way for tomorrow’s smartphones. In 1993 Caltech’s Touchstone Delta—one of the most powerful supercomputers in the world—had a peak speed of 32 gigaflops; today the Graphics Processing Units of some smartphones can reach about 75 gigaflops.
Researchers from UChicago to the Swiss Federal Institute of Technology have come together to use Mira, leading to new collaborations and advances in fields such as biology, cosmology, and more. The supercomputer also is a key resource for scientists at the Computation Institute, a joint initiative between Argonne and UChicago to advance science in many fields through innovative computational approaches.
“Each new generation of supercomputer at ALCF has enabled major breakthroughs, such as identifying the source of antibiotic resistance in a major family of bacteria and exploring how aircraft contrails can impact climate,” says Michael Papka, MS’02, PhD’09, the division director of ALCF and the deputy associate laboratory director for Computing, Environment, and Life Sciences. “Mira’s petascale computing capabilities are like hitting the accelerator on the time-to-solution for a larger and more complex class of investigations.”
Big challenges for a computing beast
Any qualified project worldwide can put Mira’s incredible abilities to use, although it’s not so easy to be qualified. For starters, in order to run on Mira, a project’s software must be able to use hundreds of thousands of processors concurrently. Mira might be capable of 10 quadrillion floating-point operations per second (your typical laptop can only do 10 billion), but there are still more potential projects than hours in the day, with subjects ranging from a climate-weather modeling study to the structure of dark energy and its relationship to the expansion of the universe.
The U.S. Department of Energy uses a competitive peer process to determine who gets to work with Mira. The time is awarded in processor hours and can range from hundreds of thousands to hundreds of millions of hours depending on the size of the project.
Benoit Roux, a University of Chicago biochemistry and molecular biology professor, is one of the qualified few. Roux uses molecular dynamic simulations to discover how large protein systems change their shape, which in turn, affects their function.
Studying such transition pathways will take a long time and won’t immediately produce new therapies. But understanding how different kinds of proteins move and change shape could yield vital insights into cancer and other diseases. It could even point to ways of making small molecules bind to malfunctioning proteins, disabling them and preventing the development of cancer.
Yet Roux says a supercomputer like Mira helps science in the broadest sense, regardless of short-term practical applications.
“Gaining fundamental knowledge is a ‘plus’ in biomedical research, and it is not always clear where things will take you,” says Roux. “As long as one is investigating deep questions, not just stamp collecting or ‘crossing t's’ and ‘dotting i's,’ fundamental research has a long-lasting impact.”
A tool for collaboration
Mira also has become a means of bringing various scientists together for the benefit of everyone’s research, says Greg Voth, a UChicago chemistry professor. Voth works on Mira along with Argonne postdoctoral scholars John Grime, Christopher Knight, and Adrian Lange to develop new computational methodologies that can predict the behavior of complex systems at widely different scales.
“One example of borrowing ideas from a seemingly unrelated field of study is the work currently under way in the group to efficiently simulate highly coarse-grained models of many molecules, which involve large regions of empty space due to an implicit solvent, using methods similar to those utilized in cosmological studies,” says Voth, who also serves as a senior scientist at Argonne and a senior fellow at the Computation Institute. “This work is being done to study, for example, the behavior of the HIV virus.”
In addition, Mira has become an educational tool, advancing not only science itself but also the minds of the upcoming generation of scientists.
“Having access to a computing resource like Mira provides excellent opportunities and experience for educating up-and-coming young scientists as it forces them to think about how to properly utilize such a grand resource very early in their careers,” Voth says. “This gives them a unique perspective on how to solve challenging scientific problems and puts them in an excellent position to utilize computing hardware being imagined now for tomorrow.”
Originally published on July 8, 2013.