FUTURE IMPLICATIONS OF GPU ACCELERATION ON PRESENT CRYPTOGRAPHIC STANDARDS

Open Access
Author:
Miller, Garrett Michael
Area of Honors:
Information Sciences and Technology
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
  • Stanley G Aungst, Thesis Supervisor
  • James A Leous, Thesis Supervisor
  • Brian Harold Cameron, Honors Advisor
Keywords:
  • CPU
  • GPU
  • Cryptography
  • Passwords
  • Hashes
  • GPU Acceleration
  • RSA
  • ECC
  • MD5
  • SHA
  • WPA
  • WPA2
  • AES
  • CUDA
  • NVidia
  • ATI
  • Cloud
  • Distributed Computing
Abstract:
This document surveys a number of recent developments in the information security field pertaining to parallel computing and cryptographic security, and demonstrates the performance gains made possible through the use of parallel computing in Graphics Processing Unit (GPU) utilization frameworks such as NVidia's CUDA and ATI's Stream frameworks. The NVidia CUDA framework is leveraged in a number of real world tests comparing several modern traditional central processing units (CPUs) and GPUs in the same cryptographic applications. Additional topics relevant to accelerating password cracking and cryptography are also examined, such as rainbow tables and solid state drives (SSDs), as well as cloud and distributed computing. Finally, the performance enhancements afforded by GPU parallel computing are compared against modern government and commercial cryptographic standards, and recommendations are made for retaining information security in the face of such dramatic performance increases.