Low-Power Hypercube Divided Memory FFT Engine Using 3D Integration

Abstract

In this article we demonstrate a floating point FFT processor that leverages both 3D integration and a unique hypercube memory division scheme to reduce the power consumption of a 1024 point FFT down to 4.227 μJ . The hypercube memory division scheme lowers the energy per memory access by 59.2% and increases the total required area by 16.8%. The use of 3D integration reduces the logic power by 5.2%. We describe the tool flow required to realize the 3D implementation and perform a thermal analysis of it.

Document Details

Document Type
Pub Defense Publication
Publication Date
Nov 01, 2010
Source ID
10.1145/1870109.1870114

Entities

People

  • Paul D Franzon
  • Samson Melamed
  • Thorlindur Thorolfsson
  • W. Rhett Davis

Organizations

  • Defense Advanced Research Projects Agency
  • North Carolina State University

Tags

Readers

  • Electrical Engineering
  • Parallel and Distributed Computing.