Communications of the ACM
Need for Speed
University of Delaware engineer Xiaoming Li is tackling new problems in computer optimization.
Credit: Courtesy of University of Delaware
University of Delaware professor Xiaoming Li is working to improve computer optimization. Li says the computer world has evolved from single-core to many-core processors as the main computation for both everyday computer applications and high-performance computing, and the main performance issue now is to help a multitude of threads to optimally share resources.
"This subtle but fundamental change in the goal of compilation leads to the overhaul of performance modeling, program profiling, and the selection of compiler transformations and their parameter values in traditional compilers," he notes. "That overhaul is the essence of our project."
Traditional compilers are designed for single-thread performance found on single-core processors, but they slow down performance when applied to the new generation of many-simple-core processors, which can run hundreds or more threads simultaneously. Li's team is studying how to adapt compiler and code optimization techniques and is developing new ways to improve resource allocation on the new architecture.
From University of Delaware
View Full Article
Abstracts Copyright © 2011 Information Inc. 
, Bethesda, Maryland, USA
No entries found