TY - GEN

T1 - Floating-point computation with just enough accuracy

AU - Dietz, Hank

AU - Dieter, Bill

AU - Fisher, Randy

AU - Chang, Kungyen

PY - 2006

Y1 - 2006

N2 - Most mathematical formulae are defined in terms of operations on real numbers, but computers can only operate on numeric values with finite precision and range. Using floating-point values as real numbers does not clearly identify the precision with which each value must be represented. Too little precision yields inaccurate results; too much wastes computational resources. The popularity of multimedia applications has made fast hardware support for low-precision floating-point arithmetic common in Digital Signal Processors (DSPs), SIMD Within A Register (SWAR) instruction set extensions for general purpose processors, and in Graphics Processing Units (GPUs). In this paper, we describe a simple approach by which the speed of these low-precision operations can be speculatively employed to meet user-specified accuracy constraints. Where the native precision(s) yield insufficient accuracy, a simple technique is used to efficiently synthesize enhanced precision using pairs of native values.

AB - Most mathematical formulae are defined in terms of operations on real numbers, but computers can only operate on numeric values with finite precision and range. Using floating-point values as real numbers does not clearly identify the precision with which each value must be represented. Too little precision yields inaccurate results; too much wastes computational resources. The popularity of multimedia applications has made fast hardware support for low-precision floating-point arithmetic common in Digital Signal Processors (DSPs), SIMD Within A Register (SWAR) instruction set extensions for general purpose processors, and in Graphics Processing Units (GPUs). In this paper, we describe a simple approach by which the speed of these low-precision operations can be speculatively employed to meet user-specified accuracy constraints. Where the native precision(s) yield insufficient accuracy, a simple technique is used to efficiently synthesize enhanced precision using pairs of native values.

UR - http://www.scopus.com/inward/record.url?scp=33746638309&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33746638309&partnerID=8YFLogxK

U2 - 10.1007/11758501_34

DO - 10.1007/11758501_34

M3 - Conference contribution

AN - SCOPUS:33746638309

SN - 3540343792

SN - 9783540343790

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 226

EP - 233

BT - Computational Science - ICCS 2006

T2 - ICCS 2006: 6th International Conference on Computational Science

Y2 - 28 May 2006 through 31 May 2006

ER -