FFTW
FFTW is a comprehensive collection of fast C routines for computing the Discrete Fourier Transform (DFT) and various special cases thereof. It is an open-source implementation of the Fast Fourier transform algorithm. It can compute transforms of real and complex-values arrays of arbitrary size and dimension. AMD FFTW includes selective kernels and routines optimized for the AMD EPYC™ processor family.
Highlights of AMD FFTW 2.0
- AMD Optimizations are enabled through configure option “–enable-amd-opt”
- Improved performance of cpy2d routine for the in-place transform of FFTW
- Enabled 256-bit SIMD kernels selection over 128-bit SIMD kernels for AMD CPU when the processor has 256-bit FPU and SIMD support
- New improved in-place transpose method targeted for very large-sized FFT. This is an optional feature for single-core execution that can be enabled by configure option “–enable-amd-trans”
- FFTW wisdom file feature (reading and writing) is extended to support multiple wisdom files corresponding to different FFT problems. This avoids overwriting of same wisdom file for different FFT problems
The package containing AMD FFTW Library binaries which include optimizations for the AMD EPYC™ processor family and documentation are available in the Downloads section below.
Source code is available on GitHub https://github.com/amd/amd-fftw.