Re: branch prediction, renaming, joins

On Wed, 03 Dec 2003 17:57:12 +0100, Terje Mathisen wrote:

> In your particular example, (d) must be independent of the outcome of 
> the if block, which means that a decent compiler would probably hoist it 
> up above instead, right?

(Does gcc hoist independent instructions above conditionals, for example?)

I would expect that a high proportion of (d) blocks *would* be dependent
on the branch, where the whole thing is the moral equivalent of:

        d(a ? b : c)

Using different notation, that's not an uncommon structure.  Of course,
that sort of thing is a perfect example of a situation where conditional
moves or predication are efficacious.

> What I'm saying is that as long as (d) depend upon (b) and/or (c),
> there's no way it _can_ be stitched back together, at least not until it
> is clear that no aliasing occured.

-- 
Andrew