Projects

Co-Designed Secure and Compressed Memory

Students: Henry Cannon

Secure memory and compressed memory systems both rely on extensions to the memory controller to implement their desired properties. This project explores the co-design of the metadata fields in order to hide the latency of compression behind the authentication cost in a secure memory.

Secure Memory for CXL

Students: Sophie Park

CXL is an emerging memory system, and designing secure memory for this platform provides significant opportunitites and challenges. This project plays with the notion of the trusted computing base in a CXL system to implement secure memory optimizations.

Concurrent Data Structures for CXL

Students: Steven Kim and Vadym Musiienko

Given the potential heterogeneity of memory devices in a CXL system, designing systems (e.g., concurrent data structures) in an architecture-aware manner remains an open problem. This project explores system-level design decisions and proposed architectural modifications to enable efficient designs.

Computability of Metadata Cache Replacement

Students: Philippe Essama

It has been demonstrated that Belady’s Min algorithm is suboptimal for metadata caching. In this project (co-advised by Prof. Mik Zlatin), we prove formally the computability of this task.

Parallel (gem5) Simulation Engine

Students: Connor Wang and Arsene Laurent

We are developing an engine that allows for events to be handled in parallel by worker threads via a dependence-aware scheduler.