After 19 Years, Kale Still Enjoys Student and Collaborator Engagement with Charm++ Workshop

Reflecting upon the fact that his Charm++ Workshop recently completed its 19th edition, Illinois CS Research Professor & Paul and Cynthia Saylor Professor Emeritus Laxmikant (Sanjay) V. Kale has come to the belief that there are two contributing factors to the event’s longevity.

Both reasons behind the success of this conference – which focuses on adaptivity in highly scalable parallel computing, recent results in adaptive runtime techniques in Charm++, and large-scale applications that rely on the runtime system – boil down to collaboration.

First, there was a clear need for production partners in other fields of science and engineering – typically biophysics or computational astronomy – to find and fix issues they encounter.

Second, the academic researchers needed to know the real-world challenges their partners had to solve to make their production-quality system – which runs on almost all parallel computers – as relevant as possible.

The Charm++ Workshop became a perfect hub for this collaborative spirit. Here, Kale and his students have grown working relationships with the people and companies who kept Charm++ relevant for so many years.

“Since the early years of my career, I’ve felt that my niche in computer science is to work on parallel programming. To do that successfully, I needed to work with excellent collaborators on the right applications,” Kale said. “Otherwise, you end up developing ideas that you consider beautiful, but when you throw these ideas over the fence and into the real world no one cares because you're not solving a relevant problem.”

When the conference started Kale already had collaborators in astronomy and biophysics.

It made sense to Kale and his collaborators to gather yearly. At the conference, Kale said, they “compare notes, give talks about what current progress looked like, and talk about what systems research was on the verge of producing next.”

Student involvement remains crucial because they work alongside Kale on these relevant projects.

They also organize the event by ensuring that the most relevant, cutting-edge research in the field that applies to Charm++ finds an outlet. This can be their own research or from others across the country.

They reach out to and schedule keynote speakers to provide access to leading experts in the field.

They also make sure industry collaborators understand what the conference is about and how it fits with their needs.

Kale’s students did the same again this year. The event came together for a virtual experience, though, due to the limitations associated with the COVID-19 pandemic.

“I've helped with the organization of the Charm++ Workshop for the past four years. This includes being the general chair for the 2019 workshop, as well as co-chair and publicity chair positions in other years,” said Jaemin Choi, a sixth year Illinois CS PhD student. “We had a lot of positive feedback for this year’s workshop, especially on the diversity and quality of the talks provided. For me, seeing that the participants get the most out of the workshop is the most satisfying moment.”

Second year PhD student Zane Fink began researching parallel computing in an undergraduate bioinformatics lab. His interest clarified when that group encountered problems that required parallel computing techniques.

“I thought I could make a real impact by pursuing research in this field,” Fink said. “I chose Illinois CS because of the breadth and depth of the department's research capabilities. Professor Kale introduced me to Charm++ in my earliest conversations with him. The application-driven development of Charm++ was important to me. Often in computer science we create solutions for problems that we ourselves invented and find interesting but have little practical application.

“At this year’s workshop, we had external talks from developers of other parallel programing systems. It was great to interact with those who help strengthen valuable connections within the broader community of high-performance computing.”

Ushering in these connections for students and working alongside of them on so many relevant research projects over the years, keeps Kale motivated.

Even if he never expected the workshop to continue for this long.

“Students are the reason I have remained a professor, because they are the creative engine behind the work,” Kale said. “Their youthful, creative energy provides new perspectives for me to consider. I enjoy working with these intelligent minds and molding them, but also being molded by them.”