Based on joint work with James Demmel, Ilie Garbacea, Qijing Huang, Sophia Shao, and many others. [an error occurred while processing this directive] Grace Dinh is a graduate student at UC Berkeley, advised by James Demmel. Her research interests include communication-avoiding algorithms, machine learning accelerator architectures, and scheduling. [an error occurred while processing this directive] Personal home page [an error occurred while processing this directive] [an error occurred while processing this directive]
[an error occurred while processing this directive]
[an error occurred while processing this directive]
[an error occurred while processing this directive]
[an error occurred while processing this directive]
PhD Candidate
[an error occurred while processing this directive]
University of California, Berkeley
[an error occurred while processing this directive]
[an error occurred while processing this directive]
[an error occurred while processing this directive]
[an error occurred while processing this directive]
Artificial Intelligence
Computer Architecture and Engineering
Theory
High-Performance Computing
[an error occurred while processing this directive]
Theoretical Methods for Optimizing Structured Array Computations
[an error occurred while processing this directive]
Structured operations on tensors, such as convolutions, dense linear algebra, and stencil operations, have become increasingly important in machine learning, numerical linear algebra, and many other domains. We present an overview of algorithms for mapping such problems onto processors (including domain-specific accelerators) in a communication-efficient manner, using lower bounds as proof of their optimality.