The Core-Awareness Principle Designing Faster Garbage Collectors For Hybrid Memories




Abstract: The advent of faster and cheaper nonvolatile memory (NVM) such as NAND flash presents system developers with unique advantages and unresolved challenges. We show that the performance of garbage-collected applications too large to fit in DRAM, but which do fit in NVM (as secondary memory) can be substantially improved. We design a new class of garbage collection algorithm called core-aware garbage collection (CAGC) that makes use of data placement to improve application performance.

Bio: Ravi Tandon is a second year Masters student in the computer science department at Princeton University. He is interested in resolving problems related to memory management when running applications on nonvolatile memory.