ACM SIGMOD, the top database conference, runs a programming contest each year to allow researchers world-wide to compete and try out how their research fares when the “rubber meets the road.” This year’s programming contest involved taking core data processing kernels that are used to analyze massive data warehouses (these are big data repositories that are now used by everyone, including your social networks, banks, healthcare provider, insurance company, to name a few examples) and make them run efficiently on modern processors. A crucial problem was how to harness the large amount of compute power that modern processors pack inside each device, which now comes in a form that is typically hard for software to harness (as the degree of hardware parallelism is high and increasing). This problem is at the heart of the Quickstep project, which aims to allow data platforms to run at “bare metal speeds” on hardware that is available today will be available in the future. (Quickstep is also a key component in a $27M grant that SRC/DARPA recently funded).
Three students from the Quickstep group, Jianqiao Zhu, Zuyu Zhang and Dylan Bacon, all Ph.D. students working with Prof. Patel, took on the 2018 ACM SIGMOD programming challenge and used techniques developed in the Quickstep project in their solution. The team took top prize at the contest holding on the leaderboard position by a comfortable margin. They were honored Wednesday afternoon at that Awards Announcement Ceremony.