High-Performance Cryptanalysis on RIVYERA and COPACOBANA Computing Systems

Tim Güneysu, Timo Kasper, Martin Novotný, Chris­tof Paar, Lars Wienbrandt, Ralf Zimmermann

"High-Performance Computing Using FPGAs", Vanderbauwhede, Benkrid (Eds.), p. 335-366, February 2013, Springer Verlag, ISBN: 978-1-4614-1790-3


Special-purpose computing platforms based on reconfigurable hardware have shown to typically exhibit a much better performance-cost ratio than off-the-shelf computers populated with general-purpose processors. In this chapter we introduce two different FPGA-based cluster architectures, called COPACOBANA and RIVYERA. These high-performance computing clusters are populated with up to 256 Xilinx Spartan or Virtex FPGAs per system and can be interconnected to form an even larger system with 2,560 FPGA per rack. In this chapter, we present a wide range of applications from the fields of cryptanalysis that have been successfully implemented on both architectures.