How did Waterloo alumnus Noel Chalmers end up a key member of the team responsible for the world’s fastest computer?
When he reflects on the winding path that led him to this position, he says he sometimes feels like a “lottery winner.” But he also credits his commitment to following his curiosity from interest to interest—even when unsure of the destination.
“I had a passion and I pursued that passion, working really hard along the way,” says Chalmers. “One experience led to another and I serendipitously found this role.”
The first twist in his journey came when, as a first-year student in aerospace engineering at Carleton University, he discovered calculus. Required to take the subject as part of his degree, Chalmers quickly became fascinated.
“I read the textbook cover to cover in the first month,” says Chalmers. “Then I then went to the library and took out more and more books. Soon, I was reading about differential equations and differential geometry. By the end of the semester, I realized I was in the wrong major.”
Chalmers transferred into applied math and went on to complete his Masters at Carleton, before applying to Waterloo for his PhD.
His time at Waterloo stands out, in part, for the brilliant people he met and worked alongside. A fellow student, working under the same PhD supervisor, had a particularly profound impact on Chalmers by introducing him to GPU computing. GPU computing gave Chalmers a way to “bring to life” the abstract mathematical concepts he was working through in his doctoral work.
“In grad school, I spent a lot of time just staring at whiteboards with piles of scrap paper in front of me, trying to conceptualize very difficult things,” says Chalmers. “I thought it was very neat that when you were done on a whiteboard, you could actually build the thing that you were analyzing. I got hooked on GPU computing very quickly.”
This discovery changed the trajectory of Chalmers’s career. Upon graduation, he completed postdoctoral work in parallel computing at both the University of Ottawa and Virginia Tech. While at Virginia Tech, he had the opportunity to attend numerous American conferences and network with people in academia and industry. One of the contacts he made at a conference worked at Advanced Micro Devices (AMD), a multinational developer of computer processors based in California, and he encouraged Chalmers to interview for a job there.
In 2018, Chalmers joined AMD’s research division. At the time, the company had been contracted by the Department of Energy to create the next generations of exascale supercomputers, including one at Oak Ridge National Lab named Frontier. Chalmers’ team acted as intermediaries between the engineers who were creating the computer and the scientists who planned to use it for research.
“It’s a nice intersection,” says Chalmers. “We learn how the hardware works at a very intimate level. And then we also understand what the software is doing and what problems the scientist are trying to solve.”
The team was also responsible for benchmarking Frontier. Chalmers himself wrote the LINPACK benchmark code used to measure Frontier’s speed. His code confirmed Frontier as the most powerful computer in the world, the first to crack the exascale barrier (capable of at least one exaflop of floating-point calculations per second). Frontier currently sits atop the top500, a list of the world’s most powerful supercomputers.
Chalmers says the unmatched power of this computer will enable scientists to increase the fidelity, resolution and accuracy of their simulations, while also giving them results many times faster—over 50 times in the case of Frontier. During the development process, Chalmers talked to scientists who planned to use Frontier to work on molecular drug discovery, full-scale cosmology simulations, combustion modelling and nuclear reactor design, among other things.
Frontier isn't the only exascale machine that Chalmers and AMD are working on. He is currently part of the team developing El Capitan, a supercomputer that will be deployed in Lawrence Livermore National Laboratory, a federal research facility in Livermore, California.
When asked what he’s most proud of about his work at AMD, Chalmers references the company’s community-oriented ethic, which prioritizes helping others and advancing the state of the industry over individual achievement.
“We view ourselves as tied into the broader computational science community,” says Chalmers. “That’s why we open source everything, including the LINPACK benchmark code I wrote. Our secret sauce is out there for anyone to see. It’s a really nice philosophy to live by.”
