The Grainger College of Engineering
Electrical & Computer Engineering

Hwu wins CGO Test of Time Award

4/18/2018 Joseph Park, ECE ILLINOIS

Published in 2008, "Program Optimization Space Pruning for a Multithreaded GPU" proposed a simple methodology for significantly diminishing the workload involved in the optimization process.

Written by Joseph Park, ECE ILLINOIS

ECE ILLINOIS Professor Wen-mei Hwu, AMD Jerry Sanders Chair of Electrical and Computer Engineering recently won the CGO Test of Time Award, also known as the Most Influential Paper Award presented by International Symposium on Code Generation and Optimization, for "Program Optimization Space Pruning for a Multithreaded GPU."

From left to right: ECE ILLINOIS alumna Teresa Johnson (BSEE '93, MSEE '95, PhD '98), Google, Program Co-Chair of CGO 2018; Michael O'Boyle, University of Edinburgh, Program Co-Chair of CGO 2018' Sara Baghsorkhi (PhD '11), Intel; Prof. Hwu; Aaron Smith, Microsoft, Sponsor Chair of CGO 2018).
From left to right: ECE ILLINOIS alumna Teresa Johnson (BSEE '93, MSEE '95, PhD '98), Google, Program Co-Chair of CGO 2018; Michael O'Boyle, University of Edinburgh, Program Co-Chair of CGO 2018' Sara Baghsorkhi (PhD '11), Intel; Prof. Hwu; Aaron Smith, Microsoft, Sponsor Chair of CGO 2018).

Originally published in the CGO proceedings in April 2008, the paper was co-authored by several students, now alumni, including Shane Ryoo (BSEE '00, MSEE '04, PhD '08), Christopher Rodrigues (MS '08, PhD '14), Samuel S. Stone (MS '07), Sara Baghsorkhi (PhD '11), Sain-Zee Ueng (MSEE '04), and John A. Stratton (BSCompE '06, MS '09, PhD '13). The researchers revealed the complexity involved in optimizing applications for highly-parallel systems and introduced one relatively simple methodology for reducing the workload involved in the optimization process by as much as 98 percent. Their work was based on the GeForce 8800 GTX using CUDA, one such highly-parallel system, and they proposed an approach for attacking the complexity of optimizing code by developing metric to judge the performance of an optimization configuration.

With the rise of inexpensive, single-chip, massively parallel platforms, more developers will be creating highly-parallel applications for platforms which will need to be optimized, emphasizing why Hwu's research is so valuable. Hwu is also affiliated with the Coordinated Science Lab at Illinois.

Share this story

This story was published April 18, 2018.