We consider prime fields of large characteristic, typically fitting on k machine words, where k is a power of 2. When the characteristic of these fields is restricted to a subclass of the generalized Fermat numbers, we show that arithmetic operations in such fields offer attractive performance both in terms of algebraic complexity and parallelism. In particular, these operations can be vectorized, leading to efficient implementation of fast Fourier transforms on graphics processing units.

We report on a multi-threaded implementation of Fast Fourier Transforms over generalized Fermat prime fields. This work extends a previous study realized on graphics processing units to multi-core processors. In this new context, we overcome the less fine control of hardware resources by successively using FFT in support of the multiplication in those fields. We obtain favorable speedup factors (up to 6.9x on a 6-core, 12 threads node, and 4.3x on a 4-core, 8 threads node) of our parallel implementation compared to the serial implementation for the overall application thanks to the low memory footprint and the sharp control of arithmetic instructions of our implementation of generalized Fermat prime fields.

The CUDA Modular Polynomial (CUMODP) Library implements arithmetic operations for dense matrices and dense polynomials, primarily with modular integer coefficients. Some operations are available for integer or floating point coefficients. Similar to other software libraries, like CuBLAS 1 targeting Graphics Processing Units (GPUs), CUMODP focuses on efficiency-critical routines and provides them in the form of device functions and CUDA kernels. Hence, these routines are primarily designed to offer GPU support to polynomial system solvers. A bivariate system solver is part of the library, as a proof-of-concept. Its implementation is presented in [10] and it is integrated in M aple 's Triangularize command 2 , since the release 18 of M aple .

No abstract

No abstract