Floating-point implementation of complex multiplication

Authors:
Earl E. Swartzlander, Jr.;Hani H. Saleh
Affiliations:
ECE Department, University of Texas at Austin, Austin, TX;Intel Corporation, Austin, TX
Venue:
Asilomar'09 Proceedings of the 43rd Asilomar conference on Signals, systems and computers
Year:
2009

Citing 3
Cited 1

Design of the IBM RISC System/6000 floating-point execution unit

IBM Journal of Research and Development
A procedure for implementing the fast Fourier transform on small computers

IBM Journal of Research and Development
Fused floating-point arithmetic for dsp

Fused floating-point arithmetic for dsp

Low-complexity adaptive decision-feedback equalization of MIMO channels

Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper extends the consideration of fused floating-point arithmetic to complex multiplication, an operation that is frequently used in DSP. There are two basic approaches to complex multiplication: (1) the conventional method that uses 4 real multiplies and 2 real adds and (2) Golub's method that uses 3 real multiplies and 5 real additions. This paper compares discrete implementations using individual floating-point adders and multipliers with implementations that use fused operations (fused dot product, fused add-subtract and combinations of the two).