A sample of Lake Superior agate collected in Northern Minnesota. Photo: Jonathan Larsen, Diadem Images.
It’s a little hard to explain just how powerful the University of Minnesota’s new supercomputer will be.
Comparatively speaking, the Minnesota Supercomputing Institute’s Agate cluster—named after Lake Superior agate, Minnesota’s state gemstone—will boast roughly seven times the computing power of the existing Mesabi system when it arrives on campus later this summer. Agate’s small army of general-purpose processors, graphics processing units (GPUs), and large memory nodes add up to a theoretical performance totaling about 7 quadrillion floating point operations per second (or 7 petaflops).
To Jim Wilgenbusch, PhD, director of research computing, the power of the supercomputer comes not through its speed, however, but what it makes possible. Cutting-edge supercomputers allow us to take incomprehensible quantities of data at the micro and macro scales, process it at a lightning fast rate, and reduce it all down to outputs our brains can understand.
“From my perspective, it’s really about changing the way we can view things,” Wilgenbusch said, equating it to clearing up a grainy television screen to allow the finer details of the image to show through. “These are things you just can’t see without supercomputers, things that would be totally invisible to us.”
Over 900 principal investigators currently use MSI’s supercomputers to conduct some portion of their research, according to Jim Ferguson, assistant director for user services. As new research applications for supercomputers continue to emerge, Agate’s capabilities will help MSI meet a growing demand among the University research community.
“Faculty start looking for high-performance computing when they can no longer use their laptop or desktop computer to get their research done,” Ferguson said, recalling one example where a researcher’s office computer left to run all night would still be working on the task when she arrived the next morning. “The additional computing power that comes with Agate is getting spread out among more people who are requiring this type of powerful computing.”
Researchers who are already running projects on MSI’s existing systems will be able to easily move resources onto the new cluster, Ferguson added. In some cases, new optimizations will be available to help them get the most out of the new system.
In tandem with the new system’s arrival, and in response to the ever-growing deluge of research data being generated across the University, MSI is introducing 15 petabytes (roughly 15.7 million gigabytes) of storage. The upgrade is crucial, as research projects ranging in scale from the genomic to the geospatial generate massive quantities of highly detailed data at a more rapid pace than ever before.
“The instruments used to collect these types of data are getting amazingly sophisticated,” Wilgenbusch said. “We’re trying to make sure our systems can keep pace with our ability to generate new data.”
Agate Knows No (Disciplinary) Bounds
Once it’s up and running, Agate stands to supercharge research across a wide variety of academic fields, from the computational sides of chemistry and biology to the socioeconomic research at the Institute for Social Research and Data Innovation to the image processing needed in areas like microscopy and MRIs of the brain.
Steven Friedenberg, PhD, DVM, assistant professor of veterinary clinical sciences in the College of Veterinary Medicine, anticipates the ways Agate will accelerate his research, which involves whole-genome sequencing of companion animals as well as the study of genetic variants and rearrangements on many of the key genes in the immune system.
“Because our datasets are so large and technically complex, we face challenges on a daily basis in terms of how fast we can process and analyze our data, and in terms of how much space we have to store our analyses,” Friedenberg said. “We are extremely excited about the opportunity to take advantage of MSI’s new system, which will dramatically reduce our run times, expand our ability to study different diseases in more animals, and essentially make discoveries at a much faster pace.”
One particular strength of the new system will be in the emerging area of machine learning, a type of artificial intelligence used across many academic disciplines where systems learn from data, identify patterns, and make decisions with relatively little human input. Agate will also include a number of single-precision GPUs, which were chosen in part for their exceptional ability to accelerate this type of research.
An Investment in UMN Research
In addition to accelerating specific research projects, Agate also represents a broader investment in the University’s research enterprise. Having high-performance computing capabilities on campus can help faculty compete for sponsored research funding and become appealing collaborators on multi-institution grants. The supercomputer also serves as a nexus for public-private collaborations by removing practical barriers to data sharing and by providing a platform for the application of advanced workflows and algorithms to solve real-world problems.
Agate’s capabilities will bolster core research functions across the institution, giving facilities such as the University Imaging Centers the ability to generate data in a way they never have been able to before. The cutting-edge system will also open up more dedicated computing opportunities for laboratories and centers around campus that require a higher degree of high-performance computing than the standard allocation provides.
“We talk about Agate in terms of both capacity and capability,” Wilgenbusch said. “With this acquisition, we’re certainly expanding in both directions.”
Agate by the Numbers
- 7 petaflops of theoretical computing power
- 770 AMD general processors
- 264 NVIDIA graphics processing units
- 15 new petabytes of storage
- 165-square-foot physical footprint