This is the mail archive of the gcc-patches@gcc.gnu.org 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: [patch] PR18716 fix - adding a gfortran testcase question


On Thursday 02 December 2004 14:30, Dorit Naishlos wrote:
> I committed the patch, but I wasn't sure how/where to add the new gfortran
> testcase?
>
> Shall I create a new 'vect' directory under gfortran.dg, like the one under
> gcc.dg, and copy over tree-vect.h and vect.exp from gcc.dg/vect?

Sounds ok to me. You'll have to figure out how you're going to deal with 
tree-vect.h, including it directly obviously isn't an option! Fortran isn't 
preprocessed by default, although you can enable it.

It should be possible to just do this test once, then skip all the vector 
execute tests if it fails. I don't know if the infrastructure for this exists 
though.

> Is this how it should look like:
>
> ! { gd-do compile }
> ! { dg-require-effective-target vect_float }
> subroutine test(n, v)
>   implicit none
>   integer :: n, i
>   real(4) :: v(n)
>
>   do i = 1, n
>     v(i) = 0
>   end do
>
>   return
> end
> ! { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } }

Yes, that looks about right. In the future you should be able to use the F2003 
C bindings to get "float" type instead of hardcoding real(4).  That hasn't 
been implemented yet though.

Paul


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