Due to the great processing power available on today’s Graphics Processing Units (GPU), we studied the suitability of mapping statistical testing algorithms to this specialized hardware. Out of the testing algorithms proposed by the National Institute of Standards and Technology (NIST), only some were suitable for implementation on GPU, due to the computational format and restrictions of the hardware. Experimental results show a significant increase in performance; very good acceleration was obtained especially for large amounts of input data. We estimate that both the performance and the categories of testing algorithms suitable for implementation will increase over time, as new GPU generations are developed and made available. In this light, using General Purpose computing on Graphics Processing Units (GPGPU) for testing sequences of random numbers is a viable and promising option for future research.
[1]
Elaine B. Barker,et al.
A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications
,
2000
.
[2]
John Owens,et al.
Streaming architectures and technology trends
,
2005,
SIGGRAPH Courses.
[3]
Mark J. Harris.
Mapping computational concepts to GPUs
,
2005,
SIGGRAPH Courses.
[4]
Dawid Pajak.
General-Purpose Computation Using Graphics Hardware for Fast HDR Image Processing
,
2007
.
[5]
Ken Umeno,et al.
Corrections of the NIST Statistical Test Suite for Randomness
,
2004,
IACR Cryptol. ePrint Arch..