Inaugural Banerjee Lecture on High Performance Computing Responds to its Namesake’s Constant Quest for New Knowledge | Computing
When Sanchita Banerjee Saxena followed Illinois Computer Science Department Head Nancy M. Amato to the podium on Thursday, April 15 at the auditorium of the National Center for Supercomputing Applications, it marked the end of a process nearly five years to honor Saxena’s father and Illinois CS alumnus Utpal Banerjee (MS ’76, PhD ’79) who died in 2017.
The inaugural Utpal Banerjee Distinguished Lecture in High Performance Computing was originally scheduled to take place almost two years ago to the day – before the COVID-19 pandemic.
As the auditorium filled to capacity, Saxena relished the moment of introducing her father to students who didn’t know him and to longtime colleagues and friends who knew him quite well. One of those friends and colleagues – Illinois CS Professor Emeritus and Intel Fellow David J. Kuck – then delivered the talk to a room full of engaged and enthusiastic minds.
This scene produced the exact meaning that Saxena had envisioned when she first discussed the idea of this annual conference with department leadership.
“It was so great to be here, seemingly all of a sudden considering it’s been two years of waiting and even more planning,” said Saxena, executive director of the Institute for South Asia Studies at the University of California at Berkeley. “Being here with my two children, Rohil and Sanil – Utpal’s grandchildren – and the rest of my family has been a wonderful time. I think we, like so many others, missed those kinds of interactions and got a little desperate about it.
“There was an undeniable energy in the room seeing everyone there. I loved it. My dad would have loved that. It was great.”
The conference itself honored Banerjee’s enduring success, which included a nearly 20-year career at Intel in the Software Solutions Group, where he developed techniques to improve the performance of multi-core processors.
When he came to Illinois CS for his PhD, Banerjee developed a strategy to automatically analyze a loop and determine if it could be run in parallel. The success of this notion – introduced to academia with the help of Kuck, his thesis supervisor – has been so immense that it remains widely known as the “Banerjee test”.
Current Illinois computer science professor Lawrence Rauchwerger connected with Kuck through his student work at the Center for Supercomputing Research and Development (CSRD) – which Kuck held as director. Also, Rauchwerger’s doctoral advisor was current Illinois CS Professor Emeritus David Padua – who was a student of Kuck and an academic brother of Utpal Banerjee.
Rachwerger introduced Kuck by linking his work to the significance of Banerjee’s.
“It’s only natural, given that we’re here at the University of Illinois, that we can use this conference to explain the profound impact that David Kuck and Utpal Banerjee have had on computing,” Rauchwerger said. “When they worked together to solve a problem with loops to make them faster, this analysis became known as the ‘Banerjee test’ and was widely used in compiler development.
“It’s still used today because, in computing, we know speed is important.”
Rauchwerger also said that Bajerjee’s work was “deep but understated, just like the man”.
This description coincided well with other complementary words.
Padova said the lecturer is “one of the most powerful voices of his generation – the generation that created the discipline.”
It quickly became apparent how much impact Banerjee and Kuck had had on their corner of IT.
“One of the things that this conference reminded me of and made me consider more was about my relationship with students,” Kuck said. “Of course, some students take different directions after graduation, and it’s easy to lose touch. But some students become lifelong friends and colleagues.
“Utpal was certainly a lifelong friend and colleague. He has also made an amazing contribution to the study of theoretical computer science here at Illinois. Since his contribution is still widely used, it’s almost as if he is always with us all the time.
Listening to Kuck deliver his talk – titled “Architecture and SW Complexity as Optimization Tool Drivers” – Saxena couldn’t help but feel inspired. She also took some time to think about what her father’s reaction would have been to hearing her mentor give that speech.
“When we were talking about my dad’s humility, it wasn’t just some funny personality trait. It was genuine. He used to say, ‘Actually, I don’t know anything. I so much to learn. It wasn’t just self-mockery, it was what he believed – and it stemmed from such respect for the amount of knowledge in the world,” Saxena said.
Banerjee’s humble nature and obvious impact on IT was a perfect fit for this conference, according to Amato.
“When Utpal passed away in 2017, his family explained that they were only beginning to understand the impact this Illinois CS alum was having on high-performance computing,” Amato said. “As a result, we created this series of lectures in his name to focus on high performance computing.
“The annual conference will remind people of the importance of high-performance computing, and it will continue the tradition and the incredible record of this department in the field by inviting others to explain their work.”
The fact that this conference will now exist year after year has special meaning for Saxena.
“One of the things that struck me, speaking with Nancy, is how permanent this conference is,” Saxena said. “We talked about ideas for how this will continue, and maybe add new formats or types of speakers from academia. There are a lot of different directions it could take, but the important thing is that it lives up to its name for as long as it exists.
“Ultimately, my father wanted to support new ideas through exciting research collaborations, and that’s what this conference represents.”