The overhead of cgo seems like it would not be an issue if you could pass in an array of vectors and moved the outer loop into the procedure instead of calling the procedure for each vector. This may only be viable if there is a low overhead way to pass arrays back and forth between go and C. I'm no expert but a quick google makes me think this is possible.
I get that hand rolling assembly is fun but as a theoretical future maintainer I would much prefer C to ASM.
I get that hand rolling assembly is fun but as a theoretical future maintainer I would much prefer C to ASM.