Emacs + Windows + utf-8

[Ale]

Hi, I use emacs both in linux and windows (native build).
I'm using a not so old emacs23 snapshot on windows (going to update
soon), and emacs 23.1 on linux.
Editing files that contain utf-8 characters like
(s¹i, s²j) ∈ S ⇒ (s¹h, s²k) ∈ S ∀ s¹h ∈ ε*(s¹i), s²k ∈ ε*(s²j)
is ok on linux (both terminal and graphical), however, Emacs feels dog-
slow  on Windows. For example, when the cursor is sitting on the line
containing those chars, if I press and hold 'a', I will see a change
in the buffer when 20 characters have already been inserted.
My setup is almost the same in windows and linux - actually the lisp
stuff I load on windows is a proper subset of the stuff I load on
linux.
Any clue why it feels so sluggish?
__
Alessandro