We are a group of computer scientists led by Prof. Abhishek Bhattacharjee.
The evidently rising heterogeneity demands increase in system utilization and performance. At the Bhattacharjee lab, we look at changing the way people have been thinking about heterogenous systems architecture traditionally and envision building an accelerator-centric system stack. We study systems ranging from data center servers to implantable brain devices and are motivated to build an end-to-end pipeline for brain-data processing.
Broadly our work in classic systems research and Brain Computer Interfaces (BCI) could be seen as follows:
Classic Systems Research: With the ever increasing heterogeneity , we ask fundamental questions such as the shift of control plane to the accelerator, the shifting of autonomy in a heterogenous system from the CPU to the accelerator, the nature of an accelerator-oriented system stack and so on. Topics that we are researching currently include asynchronous interfaces, processor-accelerator integration and communication metatheory, shared TLBs and IoT device support for virtual memory. We also work on the borders of formalization theory and integrating it with classic systems research.
BCI Research: We look at mid and low-level issues related to Brain Computer Interfaces (BCIs) and their implementation. Topics that we work on include BCI modelling, BCI storage, BCI query system and BCI verification.