Communications of the ACM
Dirac Testbed Reveals How Applications Are Written
The Dirac testbed is being put through its paces to determine which scientific codes might benefit most from being adapted to run on graphics processing units.
Credit: Margie Wylie / Lawrence Berkeley National Laboratory
Graphics processing units (GPUs) are increasingly being used in high-performance computing, but "the question is whether GPUs offer an effective solution for a broad scientific workload or for a more limited class of computations," says the Lawrence Berkeley National Laboratory's Katherine Yelick. The National Energy Research Scientific Computing Center (NERSC) and Berkeley Lab researchers recently launched Dirac, a general-purpose GPU computing testbed, to test GPUs' capabilities.
"The [U.S. Energy Department] offered funds to buy a system to study how our community writes its applications, in contrast to the typical NERSC system that is intended primarily for running them," says Berkeley Lab researcher Paul Hargrove. The system allows users to study GPUs in several applications and fields. There are currently about 500 different applications used throughout NERSC, according to Yelick. "Given that a GPU can execute thousands of parallel threads concurrently, we can potentially obtain significant speedups over the same application code optimized for a CPU," says NERSC researcher Jihan Kim.
NERSC's Filipe Maia is using Dirac to solve partial differential equations and conduct X-ray tomographic imaging. He notes that GPUs "can provide large increases in performance in many applications, which is often of crucial importance to test a wide range of conditions."
From Scientific Computing
View Full Article
Abstracts Copyright © 2011 Information Inc., Bethesda, Maryland, USA 
No entries found