About ECO Lab
Our research spans the entire computing stack — from systems and compilers to architectures and circuits, culminating in silicon-proven innovations through custom chip tapeouts. We develop novel techniques to advance reconfigurable dataflow systems, enabling high-performance execution across diverse machine learning workloads and beyond.
At the architectural level, we rethink the design of programmable fabrics and build specialized compiler infrastructures that efficiently map computations onto them. Our work extends into sparse tensor algebra acceleration, tackling the memory bottlenecks that limit scalability in modern AI workloads. Beyond CMOS, we actively explore emerging computing paradigms, including in-memory computing and photonic architectures, to push the limits of energy efficiency and performance.
On the systems side, we integrate our custom taped-out chips into energy-efficient sensor systems for ubiquitous computing applications. At the opposite end of the scale, our research in hyperscale systems explores performance modeling and scheduling for large-scale heterogeneous clusters.
News
Research Areas
Computer Architecture
Heterogeneous multi-cores, processor customization, reconfigurable computing (CGRAs)
Compiler & OS
Program analysis, timing analysis, runtime resource management and scheduling
Design Automation
Application-to-architecture synthesis, ML at the edge, IoT and edge computing
Lead PI
Tulika Mitra
Dean, School of Computing · Vice Provost (Special Projects)
Provost's Chair Professor of Computer Science, NUS
Open Positions
We are recruiting research fellows, PhD students, and research assistants for the following projects:
- MOE Tier-3 Programme GAIA — Green AI
- NRF-CRP PACE — IoT Edge Computing
Interested candidates are encouraged to email tulika@comp.nus.edu.sg.