GFFT 0.2

GFFT is generic simple and efficient Fast Fourier Transforms (FFT) implementation using policy-driven design and template metaprogramming.<br /><br />This new implementation of Fast Fourier transforms rose during study of the classical Cooley-Tukey algorithm and from an idea to assume the transform length N be a compile-time static constant, which is equal to 2^P. This assumption makes possible to implement some compile-time calculations needed for FFT such as sine and cosine functions. <br /><br />Basic features of GFFT:<br /><br /> Transforms in-place<br /> Complex FFT of power-of-two length<br /> Single real function FFT of power-of-two length<br /> High and cache-independent performance<br /> No additional data is stored<br /> You can use all available RAM for your transformed data<br /> One-step transform.<br /> Many known FFT implementation perform to steps: initialization and transform. Initialization for a given length is usually computationally expensive, while transform is fast. GFFT needs only to create an object instance that includes FFT-algorithm, but no additional data or computation.<br /> OpenMP parallelization<br />

Requirements:	No special requirements
Platforms:	*nix, Linux
Keyword:	Algorithm,  Assume,  Assumption,  Calculations,  Compiletime,  Constant,  Cosine,  Equal,  Functions,  Implement,  Implementation,  Length,  Makes,  Needed,  Policydriven,  Static,  Transform
Users rating:	0/10