need for instruction attributes (ia64)

[Richard Henderson]

On Thu, Jan 13, 2005 at 08:37:05AM +0100, Jan Beulich wrote:
> Perhaps (2) is the most reasonable thing, given that any construct
> loosely like
> 
> 	if (c)
> 		a = x ? y : z;
> 	else
> 		b = i ? j : k;
> 
> would benefit from .unc compares.

Indeed, though I don't have any good feeling of how often that happens.

> Now, that all said, would a basic (i.e. without the stuff discussed
> here) patch to add TImode arithmetic still be acceptable for 4.0?

If you can show that the existing support for TImode arithmetic
is broken, or regresses in performance from a previous release.
But otherwise I think we should probably wait.


r~