CS Events

Faculty Candidate Talk

Democratizing Error-Efficient Computing

 

Download as iCal file

Monday, March 02, 2020, 10:30am

 
radha-03-02-2020.jpg

Speaker: Radha Venkatagiri, University of Illinois at Urbana-Champaign

Bio

Radha is a doctoral candidate in Computer Science at the University of Illinois at Urbana-Champaign. Her research interests lie in the area of Computer Architecture and Systems. Radha’s dissertation work aims to build efficient computing systems that redefine “correctness” as producing results that are good enough to ensure an acceptable user experience. Radha’s research work has been nominated to the IBM Pat Goldberg Memorial Best Paper Award for 2019. She was among 20 people invited to participate in an exploratory workshop on error-efficient computing systems initiated by the Swiss National Science Foundation and is one of 200 young researchers in Math and Computer Science worldwide to be selected for the prestigious 2018 Heidelberg Laureate Forum. Radha was selected for the Rising Stars in EECS and the Rising Stars in Computer Architecture (RISC-A) workshops for the year 2019. Before joining the University of Illinois, Radha was a CPU/Silicon validation engineer at Intel where her work won a divisional award for key contributions in validating new industry standard CPU features. Prior to that, she worked briefly at Qualcomm on architectural verification of the Snapdragon processor.

Location : CoRE A 301

Event Type: Faculty Candidate Talk

Abstract: We live in a world were errors in computation are becoming ubiquitous and come from a widevariety of sources -- from unintentional soft errors in shrinking transistors to deliberate errorsintroduced by approximation or malicious attacks. Guaranteeing perfect functionality across awide range of future systems will be prohibitively expensive. Error-Efficient computing offers apromising solution by allowing the system to make controlled errors. Such systems can beconsidered as being error-efficient: they only prevent as many errors as they need to for anacceptable user experience. Allowing the system to make errors can lead to significant resource(time, energy, bandwidth, etc.) savings. Error-efficient computing can transform the way wedesign hardware and software to exploit new sources of compute efficiency; however, excessiveprogrammer burden and a lack of principled design methodologies have thwarted its adoption.My research addresses these limitations through foundational contributions that enable theadoption of error-efficiency as a first-class design principle by a variety of users and applicationdomains. In this talk, I will show how my work (1) enables an understanding of how errors affectprogram execution by providing a suite of automated and scalable error analysis tools, (2)demonstrates how such an understanding can be exploited to build customized error-efficiencysolutions targeted to low-cost hardware resiliency and approximate computing and (3) developsmethodologies for principled integration of error-efficiency into the software design workflow.Finally, I will discuss future research avenues in error-efficient computing with multidisciplinaryimplications in core disciplines (programming languages, software engineering,hardware design, systems) and emerging application areas (AI, VR, robotics, edge computing).

Contact  Faculty Host: Srinivas Narayana