Skip to content

[WIP] An attempt to write a minimalistic FFT library using Sande-Tukey and COBRA algorithms in pure C. No vectorized finalization supported, at least for now.

License

Notifications You must be signed in to change notification settings

kkrutkowski/nanoFFT

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

51 Commits
 
 
 
 
 
 

Repository files navigation

nanoFFT

An attempt to write a minimalistic FFT library using Sande-Tukey and COBRA algorithms in pure C. Only the AVX2-FMA single precision variant supports full vectorization. Partial vectorization is supported for AVX and FMA instruction sets in single and double precision.

The provided benchmarks compare performance with some of the popular FFT libraries for AMD Ryzen 4600h CPU with AVX2 and FMA sets enabled. Generally, the library tends to approximately match FFTW3f in measure mode for the scalar variant, however, its relative performance rapidly diminishes with increasing vector length.

Bench_1 Bench_2

About

[WIP] An attempt to write a minimalistic FFT library using Sande-Tukey and COBRA algorithms in pure C. No vectorized finalization supported, at least for now.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published