The Rutgers Computer Science Department is pleased to announce that Assistant Professor Sudarsun Kannan has received the National Science Foundation (NSF) Career Award for Principled System Design for Coordinated Management of Disaggregated Near-Hardware Data Processing Accelerators.
Modern computing systems struggle to handle data-intensive workloads—such as machine learning, data analytics, and graph processing—using only CPUs and GPUs, which increases data movement, energy use, and costs. This project coordinates low-power computational accelerators near memory, storage, and network resources to address these challenges, forming a scalable, high-performance, and energy-efficient framework for edge, data center, and HPC environments.
This project proposes an end-to-end design that employs specialized operating system abstractions to manage accelerator resources, supported by compiler and runtime enhancements that boost parallelism and energy efficiency. By applying ideas from distributed systems, the project aims to develop task scheduling across accelerators and deliver scalability in diverse applications—from graph analytics and machine learning to key-value stores, data analytics, and HPC simulations.
This $600K award spans over five (5) years. To learn more, click here.
