Graduate Research Symposium Keynote: Accelerating Sparse-Matrix Dense-Matrix Multiplication on GPUs

Date: 
November 6, 2020 - 4:00pm to 5:00pm
Location: 
Zoom - See emails for details
Speaker: 
Peng Jiang
UI Dept of Computer Science
https://homepage.divms.uiowa.edu/~penjiang/

Sparse matrix computations are not amenable to GPU processing due to their irregular memory access patterns and limited data reuses. Previous works have focused on accelerating sparse matrix-vector (SpMV) or sparse matrix-matrix multiplication (SpGEMM). Recently, sparse-matrix dense-matrix multiplication (SpMM) has arisen in various machine learning applications. Unfortunately, previous techniques for SpMV and SpGEMM fail to achieve good performance for SpMM. In this talk, I will introduce the current research progress in accelerating SpMM on GPUs as well as its application in accelerating CNN inference with sparse convolutions.

Bio

Peng Jiang UIowa CS Assistant ProfessorDr. Peng Jiang is an assistant professor at the University of Iowa. His research interests are broadly in software systems, with a focus on compiler and runtime techniques for parallel computing and performance optimization. Before joining UIowa, he obtained his PhD at the Ohio State University where he worked on improving the performance of applications in various domains including graph algorithms, scientific simulations, data analysis, and machine learning. His work has been published in top venues in parallel computing, compiler, and machine learning areas.