Information technology tools and resources at the UW
Computing Power Drives Discovery
Cutting-edge tools give UW scientists a distinct advantage
The interview had reached a critical point. The rising-star post-doc candidate, who Chemical Engineering Professor Jim Pfaendtner hoped to attract to the UW, posed a pointed question.
“He asked how much time he’d have to spend on computer administration for his data-hungry work of simulating biochemical processes,” Pfaendtner said. “I was proud to tell him zero. At the UW, your job isn’t computer administration. Your job is to do great science.”
That was the moment the post-doc said yes.
Like Pfaendtner, scientists and students across the UW can focus on breakthrough research thanks to the combined efforts of the eScience Institute and UW-IT to create and maintain world-class scientific computing. One key element is Hyak, a shared high-performance computing cluster, backed by lolo, an enormous, scalable storage system, and the super high-speed networks that tie the whole thing together. Another important effort is focused on managing big data by broadening awareness and facilitating the use of cloud-based services such as SQLShare, a database service that makes it easier for researchers to manage and query large datasets.
“Three years ago, central support for large-scale computational science at the UW really took off,” said Chance Reschke, UW-IT Research Computing Director. Since then, this support has enabled a diverse group of UW scientists to take leadership in compute- and data-driven discovery, Reschke said. This research ranges from complex physics and astrobiology to climate science and bioinformatics.
“The UW offers amazingly strong support for scientific computing,” Pfaendtner said. “That transforms my ability to compete for funding, to publish papers in leading journals, to make an impact in the world, and to really get students excited about the work. Hyak is miles beyond what most of my peers have.”
With Hyak and the other computing support, Pfaendtner said his Ph.D. chemical engineering students can now pose a complex question, design an experiment, and have an answer in five days.
“Ten years ago, a Ph.D. student would be lucky to do one of those calculations in two years. Even that would require heroic effort—and we’d be less sure of its accuracy. Now a Ph.D. student can do 10 or even 100,” Pfaendtner said. “When I can show prospective students a second-year Ph.D. who has published two papers, has great results and cool visualizations, it’s a big deal,” he said. “UW-IT is my backbone. I wouldn’t look anything like I do today without its support.”
Using Hyak, Pfaendtner and his research group recently developed a new method to tightly couple their simulations of biomolecules on surfaces with experimental data from the labs of their collaborators. It’s the kind of basic discovery that could have long-term implications for everything from medicines to biofuels.
“To figure out the right way to do these calculations, we needed over 400 processors running simulations for over two months,” Pfaendtner said. “Five years ago, people couldn’t even dream of doing this. Having the capability drives the science—it gives us the ability to try new ideas and take risks in ways we never could before.”