Giving nalgebra the ability to use SSE/SSE2/SSE3/SSE4/AVX instructions to preform matrix/vector computations on compatible x86 platforms would be a rather good idea, considering SSE was designed for that type of processing.
It could result in small-medium amounts of speedup (Which, as this is a mathematics library, is a very important thing)
I'm willing to investigate how to do it myself, but doing it alone would be nigh impossible.
I think @sebcrozet has made sure that nalgebra vercorises nicely, but alas it is limited to high optimizatation levels, so debug builds don't benefit. See the discussion on #274 for more information!
Giving nalgebra the ability to use SSE/SSE2/SSE3/SSE4/AVX instructions to preform matrix/vector computations on compatible x86 platforms would be a rather good idea, considering SSE was designed for that type of processing. It could result in small-medium amounts of speedup (Which, as this is a mathematics library, is a very important thing) I'm willing to investigate how to do it myself, but doing it alone would be nigh impossible.