Passat Group at UIUC is looking at Programmable, Adaptable, and Scalable Systems, Architectures, and Technologies. We are currently interested in both disruptive and non-distruptive, but impactful approaches to build robust and low power computing systems. Our current projects focus on:

  • programmable stochastic computing , a computing paradigm that embraces hardware errors in architecture and design of software and hardware systems to dramatically reduce system power and cost. Projects span program analysis, architecture, and design.
  • algorithmic techniques for fault tolerance. The goal is to identify algorithmic transformations, constructs, and invariats that can be used to significantly reduce the cost of providing system reliability, especially in context of exascale systems.
  • design and architecture of energy-constrained SoCs. Focus areas include power-balanced pipelines, peak power management, energy-proportional designs, and better-than-worst-case designs.
  • low power memory systems. Our interests lie in architectural and coding-based approaches to build error resilient memories at low power cost. Current interests include chipkill memories, die-stacked memories, and embedded DRAMs.

A typical member of the PASSAT group is creative, hardworking, motivated, and ambitious. He/she has strong programming skills, with a background in circuits, CAD, or computer architecture. He/she is open to learning new skills and considers hard problems as particularly interesting. Students who do not fit the above profile are, in general, not good fits.

Passat Group is led by Dr Rakesh Kumar.


We are always looking for motivated graduate/undergraduate students to join our group. If you are a UIUC graduate/undergraduate student looking for an advisor or if you are someone interested in applying to UIUC for graduate studies, email if you want to do research in computer architecture, CAD/architecture interactions, low power design, or reliability and fault tolerance. See our research/publications pages for a sampling of our research. Please attach your CV as well.