This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC] [autovect patch] Implement vectorization hints

On Feb 27, 2005, at 3:26 AM, Steven Bosscher wrote:

the ivdeps pragma in bar() would apply to the loop in for().

This can be handled by closing pragma's location range at the end of funciton.

the pragma would apply to the second loop, because the gimplifier
will collapse and remove the first loop.

Pragma's location range can be closed in gimplifier (when loop structure is encountered).

For the other
#pragma-needing projects there has to be interaction with the parser,
and it looks like you need that, too.  We have the new C parser now,
so it is actually possible, and the right time now, to think about a
general way to have this kind of pragmas.

I know GOMP folks are waiting for new C parser because of this. But to be honest, I was prepared for the worst case scenario. I did not know where exactly new C parser would fit in Mark's GCC 4.x timetable.

I believe this patch gives us good start in terms of providing location
range facilities. I have not looked at new C parser (I looked at very
first version earlier...) but we can come up with a general place to
close location range for such pragmas. Unless you're thinking of attaching
pragma info with various constructs like loops.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]