After over 10 years, the discovery of dark silicon remains just as consequential, with repercussions extending from computer architectures to modern technology manufacturing.
Recently, Dr. Karu Sankaralingam, Mark D. Hill and David A. Wood Professor of Computer Sciences, was elevated to Fellow status by the Institute of Electrical and Electronics Engineers (IEEE). The prestigious distinction recognizes Sankaralingam “for contributions to identifying and mitigating the challenges of dark silicon,” which articulates how power efficiency fundamentally impacts microprocessor design in the 21st century.
Conferred by IEEE’s board of directors, the award is highly selective. According to the awarding organization, “The total number selected in any one year does not exceed one-tenth of one percent of the total voting Institute membership.”
For Sankaralingam, the elevation is a “humbling” recognition of the impact of his work, both on industry and the field of computer architecture. “When I joined the Department 17 years ago, there were lots of senior faculty who had all become Fellows,” says Sankaralingam. “For me to join them is kind of a weird feeling, actually.”
He credits his PhD and Masters students with providing critical support on his award-winning body of research and joins the ranks of numerous past and present members of the Computer Sciences faculty who have received the distinction in the Department’s history.
Not just research but a “tour-de-force”
Recognized foremost for the 2011 paper Dark Silicon and the End of Multicore Scaling, which the nomination text describes as “a tour-de-force of intellectually deep processor modeling, empirical measurements, and rigorous methodology,” Sankaralingam’s elevation to Fellow cements him as a leader in his field.
The paper, published at the peak of the “multicore revolution,” pointed out that multicore processors — circuits with two or more cores ostensibly designed for improved performance and reduced power consumption — would be incapable of the scalable performance they were supposed to enable.
“We were the first to observe that, while you could build increasingly advanced chips, various underlying physical phenomena meant less and less of the chip could be active at any given time,” says Sankaralingam. “The term ‘dark silicon’ refers to the silicon that cannot be used.”
A “landmark paper” in computer architecture
In addition to receiving unprecedented front-page coverage in the New York Times, Sankaralingam and co-authors Hadi Esmaeilzadeh, Emily Blem, Renée St. Amant, and Doug Burger were praised in Communications of the ACM for their “landmark paper” revealing “the unrelenting power challenge we face in the multicore era.” Blem, now a Google engineer, received her PhD from the University of Wisconsin-Madison, where she was advised by Sankaralingam in the Vertical Research Group. Burger, Technical Fellow and Corporate VP of Microsoft Research Core, also received his PhD in Computer Sciences from UW-Madison in 1998.
Ultimately, the discovery of dark silicon changed businesses, technological manufacturing, and many of the devices we rely on daily. “Industry thought that by putting more cores on a chip, we would continue improving performance,” explains Sankaralingam. “Now we know we have to be judicious in building efficient sub blocks and making sure that work is done effectively, as opposed to expanding and consuming increasing amounts of power.”
Congratulations again, Dr. Sankaralingam!