initial vectorization functionality now in mainline

[Dorit Naishlos]

> I guess so.  Need a comment that it's not 100% correct there.

ok for mainline?

thanks,
dorit

Changelog:

        * tree-vectorizer.c (vect_can_force_dr_alignment_p): Replace
        STACK_BOUDARY with PREFERRED_STACK_BOUNDARY.

testsuite/Changelog:

        * gcc.dg/vect/vect-1.c: Remove xfail from i?86-*-* x86_64-*-* due
        to usage of PREFERRED_STACK_BOUNDARY in the vectorizer.
        * gcc.dg/vect/vect-2.c: Likewise.
        * gcc.dg/vect/vect-3.c: Likewise.
        * gcc.dg/vect/vect-4.c: Likewise.
        * gcc.dg/vect/vect-5.c: Likewise.
        * gcc.dg/vect/vect-6.c: Likewise.
        * gcc.dg/vect/vect-7.c: Likewise.
        * gcc.dg/vect/vect-25.c: Likewise.
        * gcc.dg/vect/vect-31.c: Likewise.
        * gcc.dg/vect/vect-32.c: Likewise.
        * gcc.dg/vect/vect-34.c: Likewise.
        * gcc.dg/vect/vect-36.c: Likewise.
        * gcc.dg/vect/vect-all.c: Likewise.

(See attached file: diff.Aug18)



                                                                                                                                
                      Richard Henderson                                                                                         
                      <rth@redhat.com>         To:       Dorit Naishlos/Haifa/IBM@IBMIL                                         
                                               cc:       gcc-patches@gcc.gnu.org, gcc@gcc.gnu.org,                              
                      17/08/2004 20:02          mitchell@mail.codesourcery.com, Devang Patel <dpatel@apple.com>, Olga           
                                                Golovanevsky/Haifa/IBM@IBMIL, Ira Rosen/Haifa/IBM@IBMIL                         
                                               Subject:  Re: initial vectorization functionality now in mainline                
                                                                                                                                




On Tue, Aug 17, 2004 at 07:36:52PM +0300, Dorit Naishlos wrote:
> 1) Most of the xfails on i*86 and x86_64 are due to the new restriction I
> added that considers STACK_BOUNDARY. Can something better be done for
these
> platforms? (should I check against PREFERRED_STACK_BOUNDARY instead?)

We're supposed to hope, but not rely, on PREFERRED_STACK_BOUNDARY
being correct.  But I guess until someone implements forced stack
alignment, we can't really use SSE otherwise.

I guess so.  Need a comment that it's not 100% correct there.


r~

-------------- next part --------------
A non-text attachment was scrubbed...
Name: diff.Aug18
Type: application/octet-stream
Size: 10308 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20040818/299da3ee/attachment.obj>