Brighter than NYU Lagone Health’s UltraViolet and packing more firepower than Harvard University’s Cannon, Arizona State University’s Sol Supercomputer has been listed among the top performing supercomputers globally.
The Top500, an international benchmarking organization, has ranked Sol above Harvard, NYU and Johns Hopkins. In addition, IO500, another benchmarking organization, listed Sol among the top performing research and 10 node research supercomputers in the world, along with the University of Cambridge, Stanford University and Georgia Tech. Both rankings are some of the highest ever for the university, and the first time Sol was ranked since its launch last year.
“These rankings underscore the support and empowerment Sol provides to ASU researchers,” says Sally C. Morton, executive vice president of ASU Knowledge Enterprise. “This stands as a testament to ASU’s unwavering commitment to advancing knowledge and fostering groundbreaking research. Serving as a cornerstone of infrastructure, Sol propels us towards a future where new frontiers of discovery and transformative solutions flourish.”
Since 1993, the Top500 has listed the most powerful computer systems biannually by their speed, quantified by the number of floating-point operations a computer can process per second, or flops. Sol, which can process 2.272 petaflops, or 2.272 quadrillion flops, ranks highly among academic institutions, clocking in faster than Mississippi State University’s Hercules, New York University’s Greene and University of Minnesota’s Agate.
“Sol is almost 2,000 times faster or more capable than a modern laptop, and in terms of data storage, you could say about 4,000 times that,” says Douglas Jennewein, senior director of ASU Research Computing.
While standard computers use a single central processing unit (CPU) to process operations, supercomputers contain many CPUs. This allows supercomputers to draw on the combined processing power of many CPUs — clustered together as nodes — to tackle large-scale, complex problems beyond the scope of a normal machine. Sol possesses 18,000 CPU cores spanning 178 nodes.
To achieve this level of processing and computing power, ASU drew on its close relationship with Dell Technologies. As a Dell Technologies High Performance Computing and Artificial Intelligence Center of Excellence, ASU worked with the computer and software giant to design and fabricate the new supercomputer.
“Endeavoring to find answers to the world’s toughest questions requires an incredible amount of computing performance. ASU’s Sol supercomputer is designed to do just that, powered by Dell PowerEdge servers and HPC Validated Design Solutions,” says Dave Lincoln, vice president, computer systems, solutions and portfolio management, Dell Technologies. “Building on the Dell HPC and AI Center of Excellence with ASU, Sol is the latest in a series of powerful, accelerated systems that can help speed discovery with AI modeling for a wide range of areas from protein folding to weather predictions.”
Since its inception last year, Sol and other Research Computing resources have supported the research projects of approximately 20% of principal investigators across ten colleges, including the Ira A. Fulton Schools of Engineering, The College of Liberal Arts and Sciences, the College of Health Solutions and the W.P. Carey School of Business.
“There seems to be a large number of ways in which computing can assist in research. It has really expanded in exciting ways recently, and we're just trying to meet that need,” says Gil Speyer, director of the Computational Research Accelerator.
Through ASU’s Computational Research Accelerator, Speyer works with researchers across ASU to identify opportunities where high-performance computing and software solutions can bolster and optimize projects.
Supporting an array of disciplines, Sol’s capabilities have been used in the School of Geographical Sciences and Urban Planning to represent large-scale climate models; in the School of Earth and Space Exploration to determine the amount of hydrated material within Martian rocks and soils; and in the School of Computing and Augmented Intelligence to develop deep learning models to enhance PET scans for early detection of Alzheimer’s disease.
ASU researchers interested in exploring how Sol can advance their research can request access through Research Computing.
According to Jennewein, Research Computing’s resources have enabled over $170 million in research and grant funding over the past year. With Sol’s new ranking, that number is poised to grow.
“Sol is providing a foundation for our researchers to accomplish great things at a much greater scale,” says Sean Dudley, associate vice president and chief research information officer of Knowledge Enterprise. “In particular, this supercomputer is largely designed to support both the use and development of artificial intelligence. I appreciate the opportunity to provide this impressive new resource to our research community and am excited for the work this new system will support.”
Due to the rising popularity and utility of artificial intelligence (AI) platforms, Sol was purpose built to support a larger workload of AI programs. This advancement equips the university with the tools necessary to conduct research at a larger and more competitive scale, particularly as the university expands its efforts in health care.
“From imaging data analysis to genetic research, predictive health models and drug discovery, advanced computing systems will play a crucial role in supporting ASU’s new medical school,” Jennewein says.
Sol also expands the university’s ability to not only support, but connect researchers across disciplines including engineering, business, health sciences and more. By being competitive on a global scale, it raises the caliber of the research itself.
“The effect of investing in an instrument like this on our campus is that a community of practice emerges,” Speyer says. “It seems to bring people together and network them in such a way that they become, as a community, more competitive research-wise.”