As Stanford professor John Hennessy once said, “[P]arallelism and ease of use of truly parallel computers … [is] a problem that’s as hard as any that computer science has faced.” As my Ph.D. advisor, Hennessy instilled in me a desire to conquer this problem, and I have been working at it for the last 30 years. Specifically, my research focuses on parallel computer architecture, with an emphasis on shared-memory organizations and their software layers.

What does it take to produce application code that performs as close as possible to a parallel architecture’s compute or memory peak performance? This question is one that programmers of high-performance architectures contemplate regularly since using such systems efficiently can solve problems faster, or solve larger or more complex problems. 

This question fundamentally changes the approach to programming. 

Programming is no longer simply about the correct specification of an algorithm, but expands to understanding and exploiting features of the target architecture in all aspects of an application: algorithm choice, data structures and data layout, where to exploit parallelism, how to make the best use of the memory hierarchy, and how to avoid costly communication and synchronization between cooperating computations. Building applications while addressing performance and scalability concerns is difficult and frequently leads to low-level software that exposes architectural details.

I study how data and people interact. For more than a decade, I have been studying how to help humans access and manage information. While there is a lot of good work on human-computer interaction and on data visualization, much less work exists on “human-data interaction.” Why can anyone use Google to get information of interest while it is so difficult to get useful information from a structured database? The difference lies in the specificity of the request. A web search engine receives your request and tries to guess your intention. You know that it has a limited understanding of your need, and are happy to have it get you into “the zone,” from where you can explore for yourself. On the other hand, a traditional database query engine can give you complete answers to complex questions but requires that you precisely specify your query. If you make a small mistake, you are out of luck. Wouldn’t it be helpful to devise database query mechanisms that you can actually use and get reasonable results from even if you don’t ask it totally correctly? Complementarily, can the system help you ask a better question in the first place? Similar concerns also apply to the creation of a database, and helping users manage their data.

“What are computer users doing that is wasting their time?” This question guides my research. I construe computer systems research quite broadly; if I can build it, it’s a systems problem. This breadth has let me pursue questions in visualization as well as operating systems; machine learning and computer architecture; file systems, performance analysis, graph processing, databases, and numerous other areas. Some people might say I have a short attention span; I just like to claim that I have broad interests!