Architectures for Emergining Technologies
In collaboration with Dr. Chris Dwyer I am exploring the architectural implications of emerging nanotechnologies. We are specifically interested in the impact of moving to a new fabrication method that relies on self-assembly to create ever smaller devices. Our current project in this space is investigating the potential for nanoscale photonic computing using small fluorescent molecules---called chromophores---that can be organized on DNA nanostructures. The long-term goal of this project is to develop all-optical nanoscale computers.
Multicore/Multiprocessor & Distributed Systems for Data Centers
I lead a project---called cohort cores---that builds on the observation that server workload execution patterns are not completely unique across multiple requests. We seek to develop computer systems (software and hardware) that can exploit similarity across requests to improve performance and power/energy efficiency. We seek to develop the hardware and software necessary to exploit this potential by exploring software only implementations and subsequently the design of cohort cores---a data parallel architecture that can leverage workload similarity to improve server performance via cohort scheduling. This approach is applicable to web servers, databases or other transaction based systems used in information retrieval and big data analytics. This work is in collaboration with researchers from NVIDIA and UT Austin.