Accelerating lattice reduction with FPGAs

  • Authors:
  • Jérémie Detrey;Guillaume Hanrot;Xavier Pujol;Damien Stehlé

  • Affiliations:
  • LORIA, INRIA, CNRS, Nancy Université, Vandoeuvre-lès-Nancy Cedex, France;ÉNS Lyon, Université de Lyon, Laboratoire LIP, CNRS-ENSL-INRIA-UCBL, Lyon Cedex 07, France;ÉNS Lyon, Université de Lyon, Laboratoire LIP, CNRS-ENSL-INRIA-UCBL, Lyon Cedex 07, France;CNRS, Macquarie University and University of Sydney, Dpt. of Mathematics and Statistics, University of Sydney, NSW, Australia

  • Venue:
  • LATINCRYPT'10 Proceedings of the First international conference on Progress in cryptology: cryptology and information security in Latin America
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

We describe an FPGA accelerator for the Kannan-Fincke-Pohst enumeration algorithm (KFP) solving the Shortest Lattice Vector Problem (SVP). This is the first FPGA implementation of KFP specifically targeting cryptographically relevant dimensions. In order to optimize this implementation, we theoretically and experimentally study several facets of KFP, including its efficient parallelization and its underlying arithmetic. Our FPGA accelerator can be used for both solving stand-alone instances of SVP (within a hybrid CPU-FPGA compound) or myriads of smaller dimensional SVP instances arising in a BKZ-type algorithm. For devices of comparable costs, our FPGA implementation is faster than a multi-core CPU implementation by a factor around 2.12.