We were challenged by the University of Florida’s provost, with a clear standard – that everyone, from chemists and engineers to business and social sciences students, will leave the University as a knowledgeable Artificial Intelligence (AI) citizen with at least an understanding of what AI is, how to use it, and how to do so ethically. Not only would this put additional demands on our technology resources, but we were also working with areas of study that had never used a supercomputer before. Our team stepped up to the challenge and helped to put the power of AI in the hands of researchers in areas as diverse as dentistry, agriculture and precision medicine and generated immediate and impressive results.
HiperGator3 is the high-performance computing cluster run by the team at the University of Florida Information Technology Research Computing center. The system is powered by nearly four hundred Lenovo ThinkSystem SR645 servers with AMD EPYC™ processors supporting over 40,000 processing cores as well as over 600 GPU cards. The cluster enables students and researchers to leverage the power of AI on large data sets to extrapolate insights and apply AI to unique niche areas previously not touched by supercomputing. A few examples:
- Researchers in the College of Dentistry developed a technique using x-rays to identify the precise location of where nerve canals are in a patient’s jaw when installing dental implants. Precision is key because drilling too close may cause permanent damage to the patient. Historically, identifying these nerve canals had been done by the human eye. But with the power of machine learning and image recognition, the nerve canals can be identified quickly and more accurately, increasing the procedure’s success rate and eliminating errors.
- The state of Florida has a significant agricultural economy, making crop health imperative. Collecting satellite and drone images results in massive amounts of crop data that farmers could not possibly sift through – and even if they could, they wouldn’t be fast enough to take action to resolve issues to prevent crop loss. With the speed and power of HiPerGator3 AI, researchers at College of Agriculture and Life Sciences can run algorithms to review all the image data to quickly detect specific problem areas in the crop fields allowing farmers to address issues immediately.
- College of Medicine and hospital researchers in the state of Florida are on the cutting edge of precision and personalized medicine. Among these institutions is an abundance of health and medical data where identifying patterns in illnesses, treatments and demographics is critically important. The faster a pattern is identified, the faster medical professionals can deal with it. The researchers at the University of Florida have developed a methodology using AI to gather information and identify patterns. This information is shared with doctors and nurses across the state as part of the One Florida Data Trust to help make more informed diagnostic decisions for their patients.
HiperGator3’s compute power spans beyond student research initiatives. It has a direct, positive impact on communities and citizens in the state of Florida. The University of Florida belongs to the Sunshine State Education & Research Computing Alliance (SSERCA), which is a part of the Florida Lambda Rail academic regional network provider. This group of Florida’s public and private universities missioned to further the development of a state-wide computational science infrastructure of advanced scientific computing, communication, and education resources by promoting cooperation between Florida’s universities. Researchers are able share information across their organizations.
Today, over 40 classes across the academic spectrum contain an AI component. Additionally, a group of UF students established the GAITOR Club, with a mission to “democratize AI at the University of Florida”. These students leverage HiperGator3 to run AI workloads to help non-technical faculty members run programs to extract insights from a data set.
HiperGator3 is enabling AI workloads, empowering researchers, and aiding Floridians with its vast computational abilities. As the university looks to incrementally release HiperGator4, it can only be expected that it will continue to be a driving force behind revolutionary AI-enabled discoveries.
For more information on Lenovo’s high performance computing solutions visit this page.
Erik holds a PhD in Computational Chemistry and has over 25 years of research experience. Since 2011, he has worked on building research computing infrastructure at the University of Florida. He is Senior Director of Research Computing in University of Florida Information Technology.