Re: [patch] Loop distribution for single nested loops

[Zdenek Dvorak <rakdver at kam dot mff dot cuni dot cz>]

Hello,

> On 10/3/07, Zdenek Dvorak <rakdver@kam.mff.cuni.cz> wrote:
> 
> > actually, unless it enables some other optimizations,
> > loop distribution makes the code worse, by introducing
> > an overhead of having several loops.
> 
> Right, but we are going to enable vectorization at -O3, aren't we?  I
> was thinking about that.

yes, but even then in many cases loop distribution does not help
vectorization (either because the created loops still are not
vectorizable, or because the original loop was vectorizable even without
the loop distribution); i.e., vectorizer has to somehow drive the loop
distribution, telling it which loops need to be split.

Zdenek