> Yeah. Like Tami says, this is what the instruction does.
>
> I think all three definitions are equivalent: the extend/operate/truncate
> one, the ?: one above, and the "max - min" one. Probably just personal
> preference as to which seems more natural.
Decided to switch to using the ?: one as it makes more sense
for unsigned types.
> It would be good to document what the parameters mean (except VINFO,
> which is obvious).
Documentation added for vect_recog_absolute_difference.
> I think this should instead be:
>
> if (TYPE_PRECISION (unprom_diff.type) != TYPE_PRECISION (abs_type)
> && TYPE_UNSIGNED (unprom_diff.type)
> && TYPE_UNSIGNED (abs_type))
> return false;
>
> ....
>
> I think the code would be easier to follow if it used vect_widened_op_tree
> first, and only considered the unextended case on failure.
Implemented Richard's suggested changes to vect_recog_absolute_difference.
> Minor formatting nit: GCC style is to indent braces by two spaces
> further than an "if":
>
> if (...)
> {
> ...
> }
Adopted this style.
New patch will be in the reply.
