Since #avx does not provide a dot-product on 4 double-precision floating-point numbers, below are some replacements.
Considering the dot-product can be used in many ways such as matrix-multiplication, matrix-vector-multiplication etc, this is a pity.

But hold your horses! Before typing check out what #gcc does on my naive implementation of a dot product: http://goo.gl/ZuWPsQ ; pay attention the following: first, I compiled with -Os (minimal size) and second, the placements of the parentheses does matter.

(View on Google+)

Posted in gplus.