This comment:

    ; ------------------------------------------------------------
    ; Entering the function
    ; ------------------------------------------------------------


is a little useless; it's it quite obvious that the start of the function/subroutine is right there.

However, the rest of your comments are very detailed and very well placed.

<hr>

Don't use `dec`; using `sub REG, 1` is faster.

`dec` runs in 1 CPU cycle. `sub REG, 1` runs in half a CPU cycle.

Source: [this post](http://codereview.stackexchange.com/questions/75361/countdown-program-in-x86-nasm#comment137057_75371)

<hr>

Other than that, I'm not too familiar with `xmm` registers and the instructions that go with them.

However, as I look into them, if any improvements pop into my head, I'll update my post.