Yeung Receives National Science Foundation Award
Professor Donald Yeung (ECE/CS) is the principal investigator for a new National Science Foundation award for his project entitled, "Parallelization and Memory System Techniques for Effectively Utilizing Heterogeneous Microprocessors."
The three-year, $400K grant is being funded through the National Science Foundation’s Software and Hardware Foundations (SHF) program within the Division of Computing and Communication Foundations (CCF).
Recently, CPU manufacturers have begun integrating general-purpose processors (CPUs) and graphics processing units (GPUs) on the same chip. Such heterogeneous microprocessors provide extremely low-cost communication between the CPU and GPU, enabling new ways of partitioning programs across the different types of cores. This will permit many more programs (not just those with massive amounts of parallelism) to benefit from GPUs. Dr. Yeung's project will investigate techniques for maximizing the applicability of heterogeneous microprocessors.
Major research directions include new parallelization techniques for making use of both CPUs and GPUs simultaneously, new cache coherence mechanisms to facilitate bulk producer-consumer communication between different types of cores, and adaptive memory address mapping schemes to support both CPU and GPU memory access patterns efficiently.
Published September 20, 2016