2.95, x86: severe performance problems with short arithmetic

[Jeffrey A Law]

  > Does this include something like this?
  > 
  > 	movzbw %dl, %ax
  > 	addl %eax, %esi
  > 
  > It certainly sounds like it, and would explain why such a tiny
  > difference turns into a 2x performance loss.
I believe so, yes.  However, if the destination of such an insn is not inside
a strict_low_part, we can probably turn the first instruction into a
movzbl.  That would avoid the partial register stall between the movzb/add,
but could potentially introduce a stall with an earlier insn.

jeff