LSU Researchers Awarded $1.5 Million to Advance Worldwide Access to High-Performance Computing

LSU researchers have been awarded a $1.5 million Pathways to Enable Open-Source Ecosystems, or POSE, grant from the National Science Foundation to advance the ecosystem surrounding HPX, a software library developed at LSU. Their goal is a fast and energy-efficient solution to increase worldwide access to high-performance computing. Lack of access and performance currently limits the success and usefulness of emerging technologies, such as AI and machine learning, and computational modeling in almost every field. 

While computation plays an ever-growing role is research and academic disciplines, access to high-performance computing and the talent and technology to control how it is done are markedly uneven across the world. The underlying data architectures are often disparate, making collaboration more difficult, and current capabilities are immensely hungry for energy. 

“High-performance computing is actually very much an energy problem,” said Hartmut Kaiser, who leads the LSU STE||AR Group that develops HPX and is a professor of computer science with a joint appointment in the LSU College of Engineering and the LSU Center for Computation & Technology, or CCT. “The world’s AI applications already consume more energy than we can sustain, about 47 gigawatts. This means half of the U.S. nuclear power plants work only for AI, and the projection is that this will double in the next year.” 

Kaiser and his team are working to solve these problems by creating an open-source infrastructure for high-performance computing that consumes much less energy and covers everything from the software layers down to the hardware architecture. 

“The work on HPX has grown over the last 15 years into a worldwide open-source project with contributors from all over the world,” Kaiser said. “That’s why it made sense to first establish the LSU STE||AR Group as a virtual organization and now transition the development of HPX to a foundation model, so we can more easily work with industry and other organizations that are willing to contribute, donate and pay for software development work.” 

The non-profit HPX Foundation (whose income will support students and research at LSU) will oversee workforce development efforts in collaboration with Southern University, a historically black university in Baton Rouge. In 2022, LSU and Southern University signed the A&M Agenda to increase collaboration and expand the positive and collective impact of the two land-grant institutions.  

“Part of this grant is that we want to foster a diverse community of users and developers of HPX and high-performance computing across the U.S.,” said Mohammad Tohid, a research scientist at CCT and principal investigator on the project, who will help develop new courses in high-performing computing to be taught at Southern University and online. “We’re also collaborating with the University of Washington and Professor Andrew Lumsdaine to make our solutions available to everyone through the cloud.”  

The team is already supporting two cutting-edge research areas at LSU as use cases—medical physics where HPX is used to model the cardiovascular system in the human body, and in coastal modeling of storm surge and flooding where HPX improves the infrastructure behind LSU’s nationally used Coastal Emergency Risks Assessment, or CERA. 

“We’ve worked with medical physics to simulate just one pulse in a human vascular system and that took more than half a day using a supercomputer,” Tohid said. “By leveraging HPX, we aim to enhance simulation efficiency and make significant progress toward creating a human cardiovascular digital twin.”

Professor Wayne Newhauser who leads LSU’s medical physics program is a co-principal investigator on the project. 

“Our research team focuses on finding new ways to improve outcomes for patients with life-threatening diseases—for example, we are researching how to avoid side effects in children who receive radiation therapy for brain tumors,” Newhauser said. “To do this, we need computer models to simulate all the key physical and biological steps that occur between initial cellular injury and manifestation of clinical symptoms. This project will leverage the latest computer science methods to allow us to create the first true cardiovascular digital twin, which will open new avenues of research on cancer treatments and cardiovascular disease.”  

While many high-performance computing applications have been private, isolated and not connected to the internet, the POSE project will make the technology open-source and cloud-based. This requires new levels of security, which will be enabled by co-principal investigator Professor Golden G. Richard III, director of the LSU Cyber Center. 

“It’s important for large software projects to have thorough security review and testing,” Richard said. “When you’re employing high-performance computing in internet-facing applications, security becomes even more important. That’s because anything that is open to the internet technically could be attacked by anything or anyone, including nation-states.” 

There is yet another aspect of security encompassed by the project—human security, which depends on ethical, fair and inclusive technology use and development. This is why Deborah Goldgaber, director of the LSU Ethics Institute, also joined the team as a co-principal investigator. 

“I think this team of researchers, including Hartmut and Mohammad, is particularly forward-looking,” Goldgaber said. “They see the importance of both the technical and social aspects of the project in building and sustaining a flourishing open-source community. Open-source projects have always been guided by an ethos of inclusive collaboration, and their success or failure often depends on the quality of the community that supports them.”  

HPX relies on concurrent and parallel programming models for efficiency and performance. The researchers will develop an open and inclusive HPX-based ecosystem based on two widely adopted open-source technologies: the RISC-V architecture and the LLVM/MLIR compiler.