• DocumentCode
    2976922
  • Title

    Breaking Weak 1024-bit RSA Keys with CUDA

  • Author

    Scharfglass, Kerry ; Weng, Darrin ; White, Jonathan ; Lupo, Chris

  • Author_Institution
    Comput. Sci. Dept., California Polytech. State Univ., San Luis Obispo, CA, USA
  • fYear
    2012
  • fDate
    14-16 Dec. 2012
  • Firstpage
    207
  • Lastpage
    212
  • Abstract
    An exploit involving the greatest common divisor (GCD) of RSA moduli was recently discovered [1]. This paper presents a tool that can efficiently and completely compare a large number of 1024-bit RSA public keys, and identify any keys that are susceptible to this weakness. NVIDIA´s graphics processing units (GPU) and the CUDA massively-parallel programming model are powerful tools that can be used to accelerate this tool. Our method using CUDA has a measured performance speedup of 27.5 compared to a sequential CPU implementation, making it a more practical method to compare large sets of keys. A computation for finding GCDs between 200,000 keys, i.e., approximately 20 billion comparisons, was completed in 113 minutes, the equivalent of approximately 2.9 million 1024-bit GCD comparisons per second.
  • Keywords
    graphics processing units; parallel architectures; parallel programming; performance evaluation; public key cryptography; 1024-bit RSA public keys; CUDA; GCD; GPU; NVIDIA; RSA moduli; graphics processing units; greatest common divisor; massively-parallel programming model; performance speedup; word length 1024 bit; Arrays; Graphics processing units; Instruction sets; Kernel; Matrix decomposition; Organizations; Public key; CUDA; RSA; greatest common divisor; parallel computation;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Parallel and Distributed Computing, Applications and Technologies (PDCAT), 2012 13th International Conference on
  • Conference_Location
    Beijing
  • Print_ISBN
    978-0-7695-4879-1
  • Type

    conf

  • DOI
    10.1109/PDCAT.2012.58
  • Filename
    6589265