This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [patch][gomp4] openacc loops
- From: Ilmir Usmanov <i dot usmanov at samsung dot com>
- To: Cesar Philippidis <cesar at codesourcery dot com>
- Cc: "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>, fortran at gcc dot gnu dot org, Jakub Jelinek <jakub at redhat dot com>, burnus at net-b dot de, thomas at codesourcery dot com
- Date: Tue, 17 Jun 2014 08:50:41 +0400
- Subject: Re: [patch][gomp4] openacc loops
- Authentication-results: sourceware.org; auth=none
- References: <538E538E dot 9010304 at codesourcery dot com> <538F24E2 dot 5070507 at samsung dot com> <538FEACE dot 4070703 at codesourcery dot com>
Hi Cesar!
Sorry for long delay, I was busy at the institute.
On 05.06.2014 07:58, Cesar Philippidis wrote:
I didn't see a whole lot different from your gfc_trans_oacc_loop and the
existing gfc_trans_omp_do, so I augmented the latter to handle the
openacc loop clause. It looks like the only important change is ensure
that gfc_trans_omp_do creates an OACC_LOOP stmt for the for openacc. I
know that gfc_trans_omp_do also handles the lastprivate clause, but that
shouldn't matter for the loop directive since it can't get set. Likewise
for the SIMD stuff.
Is there any specific reason why you created a new function to handle
openacc loops?
Yes, there was... I was going to support DO CONCURRENT loops.
I've replaced the asserts with sorry messages in my other patch, so it
shouldn't ICE anymore.
If this patch is OK with you, please commit it.
Now this patch looks good for me.
Thomas, could you, please, commit it?
--
Ilmir.