Speaker: Bernhard Langer (ICGA)
The majority of computer applications use numerical data types with a fixed amount of precision for their computations. Both the numerical range and precision of such types are sufficient for most use cases.
For some purposes however, such as public-key cryptography, certain numerical algorithms, or geometrical computations, the required range and precision can become arbitrarily large.
Numerical types that can handle such requirements naturally have higher memory requirements and are not natively supported by common hardware, which leads to increased computational complexity.
In this paper, we examine how basic arithmetic operations on arbitrary-precision integers can be adapted to massively parallel hardware in the form of graphics processing units, which are widely available as commodity hardware.
Apart from a detailed description of our method, we provide an evaluation of the performance characteristics of our implementation in comparison to state-of-the-art CPU libraries for arbitrary-precision integer arithmetics.