Argonne National Laboratory Mathematics and Computer Science Division
Argonne Home > MCS Division >

Publications

W. D. Gropp, "Issues in Accurate and Reliable Use of Parallel Computing in Numerical Programs," Preprint ANL/MCS-P1193-0804, August 2004. [pdf]

Parallel computing, as used in technical computing, is fundamentally about achieving higher performance than is possible with a uniprocessor. The unique issues for accurate and reliable computing are thus directly related to this desire for performance. This paper briefly covers the major types of parallel computers, the programming models used with them, and issues in the choice and implementation of algorithms.


The Office of Advanced Scientific Computing Research | UChicago Argonne LLC | Privacy & Security Notice | ContactUs