Adve’s Research on Memory Consistency Models, Influence on Hardware and Software Worlds Recognized with Kennedy Award
Illinois Computer Science Professor Sarita Adve has been named the recipient of the 2018 ACM/IEEE-CS Ken Kennedy Award by the Association for Computing Machinery and IEEE Computer Society. The award recognizes her influential work and leadership at the interface of hardware and software.
The 2018 Kennedy Award, which was announced Oct. 31, recognizes Adve “for research contributions and leadership in the development of memory consistency models for C++ and Java, for service to numerous computer science organizations, and for exceptional mentoring.”
Adve, who is the Richard T. Cheng Professor of Computer Science, said the research behind the award began in her earliest work, at the University of Wisconsin-Madison, where she earned a PhD in 1993.
“This is an area I started working on as a grad student and continue to work on. That’s quite unusual, but just happened to be a problem that was a really hard problem,” she said. “One of the reasons that it’s hard is that this problem affects both hardware and software. … It’s at the heart of the notion of correct behavior of a computer, and yet this is a part of that puzzle that has been really hard to solve.”
Adve co-developed the memory models for C++ and Java (with Hans Boehm, Bill Pugh, and others), based on her early work on data-race-free (DRF) models (with her PhD advisor, Mark Hill). The memory model specifies what value a read of a memory address will return, and is key to the correctness and performance of threaded programs, languages, compilers, and hardware. By impacting the models of the most widely used programming languages, Adve’s work has influenced the worldwide software community and hardware design.
But she notes the idea did not take hold overnight.
“We wrote the first paper in 1990,” she said. “It was very well received. But it took a long time for people to really understand our idea that a memory model should be viewed as a contract between hardware and software where software obeys certain rules (data-race-free) and hardware promises sequential consistency. That this idea struck the sweet spot between programmability and performance took over a decade to really take hold.”
More recently, with her students, Adve questioned the conventional wisdom for memory models for heterogeneous systems and showed that DRF is a superior model even for such systems.
One of the students who played a key role in that research was Matt Sinclair (PhD CS ’17), now an assistant professor at the University of Wisconsin.
“Sarita is an excellent mentor, and I've benefited tremendously from the opportunity to work with her and learn from her,” Sinclair said. “In particular, Sarita is really good at treating each person uniquely, and identifying the best way to mentor them and bring out their best. This is something I strive to emulate in my post-graduate career.”
Adve’s broader research interests are at the hardware-software interface and span the system stack from hardware to applications. She is also known for her innovations in cache coherence, hardware reliability, and power management.
Pennsylvania State University Emeritus Professor and Illinois Computer Science graduate Mary Jane Irwin (MS CS ’75, PhD ’77) nominated Adve for the award and called her "a natural and right choice," based on her teaching, mentoring, and community service, as well as the impact of her research.
"Sarita’s long-term and impactful work on designing and formally specifying memory consistency models has had profound, real-world impact on programmability and productivity in computing," Irwin wrote. "The memory consistency model lies at the heart of the semantics of any threaded software or hardware. Arguably, it has been one of the most challenging and contentious areas in concurrent hardware and software specification for many years."
The award also recognizes Adve, in part, for her extensive service to the computing community.
“To be a good researcher is not sufficient in my mind, we need to give back,” she said.
Her service includes working as the current chair of the ACM Special Interest Group on Computer Architecture (SIGARCH), where she instituted changes that have inspired new energy in the executive committee, leading to new programs in communications, research visioning, and mentoring. With colleagues, she also has made diversity and inclusion a key focus and led the creation of CARES, a committee to provide support to those who experienced harassment at SIGARCH and SIGMICRO sponsored events.
Adve also serves on the ACM Council and the DARPA ISAT study group and previously served on the board of the Computing Research Association and the NSF CISE advisory committee.
Adve is a recipient of the Anita Borg Institute Women of Vision award in innovation, the ACM SIGARCH Maurice Wilkes award, and an Alfred P. Sloan Research Fellowship. She is a Fellow of the ACM and IEEE and was named a University Scholar by the University of Illinois.
The award will be presented at SC 18: The International Conference for High Performance Computing, Networking, Storage and Analysis, in Dallas, Texas on Nov. 13.
Kuck, who is an emeritus professor, won the 2010 Kennedy Award for “pioneering contributions to compiler technology and parallel computing, the profound impact of his research on industry, and the widespread and long-lasting influence of his teaching and mentoring.” Gropp, who is both the director of the National Center for Supercomputing Applications and the Thomas M. Siebel Chair in Computer Science, was recognized with the 2016 award for “highly influential contributions to the programmability of high-performance parallel and distributed computers, and extraordinary service to the profession.”
ACM and IEEE-CS co-sponsor the Kennedy Award, which was established in 2009 to recognize substantial contributions to programmability and productivity in computing and significant community service or mentoring contributions. It was named for the late Ken Kennedy, founder of Rice University’s computer science program.