Is Intel MKL worth it for small vectors/matrices (2-4 D)

Viewed 115

I want a fast vector math implementation for a video game. Even though I'll be working with 2-4 dimensional vectors 99% of the time, I'll be doing hundreds to a few thousands vector operations every frame (on CPU), so performance is important.

I've been considering Eigen and Blaze, because the benchmarks show that they are fast enough for what I want.

However, both of them recommend using Intel MKL for even better performance.

I know Intel MKL can be very useful for large matrices, but I was wondering how much of a difference does it make for small vectors?

Has anyone ever used it for such applications?

0 Answers
Related