This is a small library of AVX/AVX2-optimized functions to process vectors of complex numbers.
The functions in this repo take advantage of Intel's intrinsic instruction sets AVX and AVX2, thus only AVX-supported CPUs have the opportunity to benefit from this repo. There are two shared libraries to choose from. If your CPU supports AVX, but not AVX2 the functions marked with "(AVX2)" in the documentation below will not work and you should use libHComplexAVX.so. Otherwise use libHComplexAVX2.so.
void Hcmul(float *dst, float *a, float *b, const int n)
Complex multiplication: Performs complex multiplication on vectors a and b, both of length n, and stores result in dst.
All vectors are on Intertwined format (see bottom of the page if unfamiliar).
void Hcmul_sep(float *dst_real, float *dst_imag, float *a, float *b, const int n);
Complex multiplication: Performs complex multiplication on vectors a and b, both of length n. The resulting real parts are stored in dst_real and the imaginary parts in dst_imag. a and b are on Intertwined format.
float Hdot(float *a, float *b, const int n)
Dot product: Performs a dot product on regular float vectors (not complex) a and b, both of length n, and returns result.
void Hmagnitude(float *dst, float *src, const int n) (AVX2)
Magnitude: Calculates the magnitude of the complex float vector src of length n and stores the result in dst. dst must be on Intertwined format.
float Hvsum(float *vec, const int n)
Vertical sum: Adds all elements of a vector vec of length n and returns the resulting sum.
[Re{z1}, Im{z1}, Re{z2}, Im{z2}, ...]