NHR PerfLab Seminar: Designing for Portable, Efficient and Explainable Performance

Symbolic picture for the article. The link opens the image in a large view.

Speaker: Tze Meng Low, Carnegie Mellon University

Title: Designing for Portable, Efficient and Explainable Performance

Date and time: Tuesday, May 10, 2:00 p.m. – 3:00 p.m.


Abstract: Modern computing platforms are becoming more complex, and more difficult to program efficiently. Existing code generators rely on search to find good implementations, but the increasing architecture complexity means searches are taking significantly longer, and the resulting code from search do not provide useful insights that can be applied to other architectures or application domains. In this talk, we present our approach of using analytical models that describe the interaction between software and hardware for tackling the issue of performance portability. Using examples from various application domains, we show that our models provide insights into why certain optimizations/design decisions are made and how the rationale behind these models can be extended to different domains and architectures.

Speaker bio:

Tze Meng Low is an Assistant Research Professor with the Department of Electrical and Computer Engineering at Carnegie Mellon University. He graduated from the University of Texas at Austin with an M.S.(C.S) in 2004, and a Ph.D. in Computer Science in 2013. His research focuses on the systematic derivation and implementation of high-performance algorithms through the use of formal methods and analytical models. His goal is to achieve performance portability across both architectures and domains by understanding and capturing the interaction between software algorithms and hardware features through analytical models so as to build better code-generators, and/or software libraries for emerging domains and architectures.