Abstract
Today’s explosive data growth has ushered a new generation of applications that transform massive, unstructured, heterogeneous data into actionable knowledge. Data is increasing exponentially in volume, velocity, variety, and complexity. On the other hand, the performance of memory systems used to store and access this data has remained almost constant throughout the years. Therefore, traditional memory systems cannot keep up with the growing demands and complexities of data-intensive applications.
In this talk, I will present my group’s research effort in optimizing the memory system performance of a variety of data-intensive applications. In particular, I will present Prodigy [HPCA 2021 Best Paper] in detail that uses a hardware-software co-designed solution to improve the memory system performance of data-indirect irregular workloads in detail. Prodigy proposes a compact, yet efficient representation of program semantics that communicates key workload information from software to hardware. Using compiler analysis and hardware prefetching, Prodigy improves the end-to-end performance of irregular workloads by more than 2.5x on CPUs. At the end, I will briefly summarize our other recent/ongoing works on optimizing other interesting data-intensive workloads.
Bio
Nishil Talati is an Assistant Research Scientist (Research Faculty) at the CSE department of University of Michigan. He earned his PhD from University of Michigan. Nishil’s research interests include computer architecture and systems software design for improving the performance of modern data-intensive workloads. His research is published at several top-tier venues including ISCA, MICRO, HPCA, ASPLOS, and others. Nishil’s work has been recognized as the 2021 HPCA best paper award, 2023 DATE and 2023 IISWC best paper honorable mentions.
