Turing Award Winner On His Pioneering Algorithms

Jack Dongarra’s dream job growing up was to teach science at a public high school in Chicago.

“I was pretty good in math and science, but I wasn’t a particularly good student,” Dongarra says, laughing.

After he graduated high school, there was only one university he wanted to attend: Chicago State. That’s because, he says, it was known for “churning out teachers.” Chicago State accepted his application, and he decided to major in mathematics.

His physics professor suggested that Dongarra apply for an internship at the Argonne National Laboratory, in Lemont, Ill., a nearby U.S. Department of Energy science and engineering research center. For 16 weeks he worked with a group of researchers designing and developing EISPACK, a package of Fortran routines that compute the eigenvalues and eigenvectors of matrices—calculations common in scientific computing.

Dongarra acknowledges he didn’t have a background in or knowledge of eigenvalues and eigenvectors—or of linear algebra—but he loved what he was doing. The experience at Argonne, he says, was transformative. He had found his passion.

“I thought it was a cool thing to do,” he says, “so I kept pursuing it.”

The IEEE Life Fellow has since made pioneering contributions to numerical algorithms and libraries for linear algebra, which allowed software to make good use of high-performance hardware. His open-source software libraries are used in just about every computer, from laptops to the world’s fastest supercomputers.

The libraries include basic linear algebra subprograms (BLAS), the linear-algebra package LAPACK, parallel virtual machines (PVMs), automatically tuned linear algebra software (ATLAS), and the high-performance conjugate gradient (HPCG) benchmark.

For his work, he was honored this year with the 2021 A.M. Turing Award from the Association for Computing Machinery. He received US $1 million as part of the award, which is known as the Nobel Prize of computing.

“When I think about previous Turing Award recipients, I’m humbled to think about what I’ve learned from their books and papers,” Dongarra says. “Their programming languages, theorems, techniques, and standards have helped me develop my algorithms.

“It’s a tremendous honor to be this year’s recipient. The award is a recognition by the computer-science community that the contributions we are making in high-performance computing are important and have an impact in the broader computer-science community and science in general.”

Dongarra didn’t end up teaching science to high school students. Instead, he became a professor of electrical engineering and computer science at the University of Tennessee in Knoxville, where he taught for 33 years. The university recently named him professor emeritus.

Entrepreneurial Spirit

After graduating from Chicago State in 1972 with a bachelor’s degree in mathematics, Dongarra went on to pursue a master’s degree in computer science at the Illinois Institute of Technology, also in Chicago. While there he worked one day a week for Argonne with the same team of researchers. After he got his degree in 1973, the lab hired him full time as a researcher.

With encouragement from his colleagues to pursue a Ph.D., he left the lab to study applied mathematics at the University of New Mexico in Albuquerque. He honed his knowledge of linear algebra there and began working out algorithms and writing software.

He returned to Argonne after getting his doctorate in 1980 and worked there as a senior scientist until 1989, when he got the opportunity to fulfill his dream of teaching.

He was offered a joint position teaching computer science at the University of Tennessee and conducting research at the nearby Oak Ridge National Laboratory which, like Argonne, is a Department of Energy facility.

“It was time for me to try out some new things,” he says. “I was ready to try my hand at academia.”

He says Oak Ridge operated in a similar way to Argonne, and the culture there was more or less the same.

“The challenge,” he says, “was becoming a university professor.”

“The Turing Award is a recognition by the computer science community that the contributions we are making in high-performance computing are important, and have an impact in the broader computer science community and science in general.”

University culture is very different from that at a government laboratory, he says, but he quickly fell into the rhythm of the academic setting.

Although he loved teaching, he says, he also was attracted to the opportunity the university gave its instructors to work on technology they are passionate about.

“You follow your own path and course of research,” he says. “I’ve prospered in that environment. I interact with smart people, I have the ability to travel around the world, and I have collaborations going on with people in many countries.

“Academia gives you this freedom to do things and not be constrained by a company’s drive or its motivation. Rather, I get to work on what motivates me. That’s why I’ve stayed in academia for so many years.”

In 1980, Dongarra worked as a senior scientist at Argonne National Laboratory, in Lemont, Ill.Jack Dongarra

Dongarra founded the university’s Innovative Computing Laboratory, whose mission is to provide tools for high-performance computing to the scientific community. He also directs the school’s Center for Information Technology Research.

He is now a distinguished researcher at Oak Ridge, which he calls “a wonderful place, with its state-of-the-art equipment and the latest computers.”

Software for Supercomputers

It was working in creative environments that led Dongarra to come up with what many describe as world-changing software libraries, which have contributed to the growth of high-performance computing in many areas including artificial intelligence, data analytics, genomics, and health care.

“The libraries we designed have basic components that are needed in many areas of science so that users can draw on those components to help them solve their computational problems,” he says. “That software is portable and efficient. It has all the attributes that we want in terms of being understandable and providing reliable results.”

He’s currently working on creating a software library for the world’s fastest supercomputer, Frontier, which recently was installed at the Oak Ridge lab. It is the first computer that can process more than 1 quintillion operations per second.

Computer-Science Recognition

Dongarra has been an IEEE member for more than 30 years.

“I enjoy interacting with the community,” he says in explaining why he continues to belong. “Also I enjoy reading IEEE Spectrum and journals that are relevant to my specific field.”

He has served as an editor for several IEEE journals including Proceedings of the IEEE, IEEE Computer Architecture Letters, and IEEE Transactions on Parallel and Distributed Systems.

Dongarra says he’s a big promoter of IEEE meetings and workshops, especially the International Conference for High Performance Computing, Networking, Storage, and Analysis, sponsored by ACM and the IEEE Computer Society, of which he is a member. He’s been attending the event every year since 1988. He has won many awards at the conference for his papers.

“That conference is really a homecoming for the high-performance computing community,” he says, “and IEEE plays a major role.”

IEEE is proud of Dongarra’s contributions to computing and has honored him over the years. In 2008 he received the first IEEE Medal of Excellence in Scalable Computing. He also received the 2020 Computer Pioneer Award, the 2013 ACM/IEEE Ken Kennedy Award, the 2011 IEEE Charles Babbage Award and the 2003 Sidney Fernbach Award.

“I’m very happy and proud to be a member of IEEE,” he says. “I think it provides a valuable service to the community.”

This article appears in the December 2022 print issue.