public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd
@ 2014-08-04 10:48 alexander.grund at mailbox dot tu-dresden.de
2014-08-04 10:48 ` [Bug fortran/62007] " alexander.grund at mailbox dot tu-dresden.de
` (8 more replies)
0 siblings, 9 replies; 10+ messages in thread
From: alexander.grund at mailbox dot tu-dresden.de @ 2014-08-04 10:48 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
Bug ID: 62007
Summary: default(none) conflicts with iteration variable in
openmp parallel loop simd
Product: gcc
Version: 4.9.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: alexander.grund at mailbox dot tu-dresden.de
If you use the "default(none)" clause in a parallel loop simd construct
gfortran will complain about having not set the iteration variable in a
private/shared clause:
" error: 'i_ct' not specified in enclosing parallel"
If you however do specify it in a private clause, gfortran will yield:
"Error: !$OMP PARALLEL DO SIMD iteration variable present on clause other than
LINEAR at (1)"
Trivial example:
!$omp parallel do simd default(none)
do i_ct = 1, 10
...
end do
If you use a "normal" parallel do both variants work.
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
@ 2014-08-04 10:48 ` alexander.grund at mailbox dot tu-dresden.de
2014-08-04 10:53 ` dominiq at lps dot ens.fr
` (7 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: alexander.grund at mailbox dot tu-dresden.de @ 2014-08-04 10:48 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
Alexander Grund <alexander.grund at mailbox dot tu-dresden.de> changed:
What |Removed |Added
----------------------------------------------------------------------------
Severity|normal |major
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
2014-08-04 10:48 ` [Bug fortran/62007] " alexander.grund at mailbox dot tu-dresden.de
@ 2014-08-04 10:53 ` dominiq at lps dot ens.fr
2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
` (6 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: dominiq at lps dot ens.fr @ 2014-08-04 10:53 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
Dominique d'Humieres <dominiq at lps dot ens.fr> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |WAITING
Last reconfirmed| |2014-08-04
Ever confirmed|0 |1
Severity|major |normal
--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Could you provide a complete test code?
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
` (2 preceding siblings ...)
2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
@ 2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
2014-08-04 11:04 ` alexander.grund at mailbox dot tu-dresden.de
` (4 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: alexander.grund at mailbox dot tu-dresden.de @ 2014-08-04 11:02 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
--- Comment #3 from Alexander Grund <alexander.grund at mailbox dot tu-dresden.de> ---
Created attachment 33233
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=33233&action=edit
Broken example
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
2014-08-04 10:48 ` [Bug fortran/62007] " alexander.grund at mailbox dot tu-dresden.de
2014-08-04 10:53 ` dominiq at lps dot ens.fr
@ 2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
` (5 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: alexander.grund at mailbox dot tu-dresden.de @ 2014-08-04 11:02 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
--- Comment #2 from Alexander Grund <alexander.grund at mailbox dot tu-dresden.de> ---
Created attachment 33232
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=33232&action=edit
Working example
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
` (3 preceding siblings ...)
2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
@ 2014-08-04 11:04 ` alexander.grund at mailbox dot tu-dresden.de
2014-08-04 12:29 ` anlauf at gmx dot de
` (3 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: alexander.grund at mailbox dot tu-dresden.de @ 2014-08-04 11:04 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
--- Comment #4 from Alexander Grund <alexander.grund at mailbox dot tu-dresden.de> ---
I attached 2 testcases. Both are the same files but with different file
endings. The .f works, the .F90 doesn't (it shows the reported behavior)
compile both with "gfortran -fopenmp {file}"
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
` (4 preceding siblings ...)
2014-08-04 11:04 ` alexander.grund at mailbox dot tu-dresden.de
@ 2014-08-04 12:29 ` anlauf at gmx dot de
2014-08-04 12:36 ` alexander.grund at mailbox dot tu-dresden.de
` (2 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: anlauf at gmx dot de @ 2014-08-04 12:29 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
Harald Anlauf <anlauf at gmx dot de> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |anlauf at gmx dot de
--- Comment #5 from Harald Anlauf <anlauf at gmx dot de> ---
(In reply to Alexander Grund from comment #4)
> I attached 2 testcases. Both are the same files but with different file
> endings. The .f works, the .F90 doesn't (it shows the reported behavior)
>
> compile both with "gfortran -fopenmp {file}"
The reason you get no error with the .f file is that it is
a Fortran fixed format source file.
If you add -ffree-form to the gfortran options, you get the
same error.
I think the problem is that the simd directive belongs to
OpenMP 4.0 and your gfortran supports only 3.1.
You'll have to wait until OpenMP 4 is supported by gfortran.
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
` (5 preceding siblings ...)
2014-08-04 12:29 ` anlauf at gmx dot de
@ 2014-08-04 12:36 ` alexander.grund at mailbox dot tu-dresden.de
2014-12-06 22:13 ` dominiq at lps dot ens.fr
2014-12-08 13:32 ` jakub at gcc dot gnu.org
8 siblings, 0 replies; 10+ messages in thread
From: alexander.grund at mailbox dot tu-dresden.de @ 2014-08-04 12:36 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
--- Comment #6 from Alexander Grund <alexander.grund at mailbox dot tu-dresden.de> ---
gfortran 4.9.1 supports OpenMP 4.0
Removing the default(none) clause removes the error message, so it certainly IS
supported.
Besides that: Even the error message mentiones "!$OMP PARALLEL DO SIMD" so it
is recognized.
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
` (6 preceding siblings ...)
2014-08-04 12:36 ` alexander.grund at mailbox dot tu-dresden.de
@ 2014-12-06 22:13 ` dominiq at lps dot ens.fr
2014-12-08 13:32 ` jakub at gcc dot gnu.org
8 siblings, 0 replies; 10+ messages in thread
From: dominiq at lps dot ens.fr @ 2014-12-06 22:13 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
Dominique d'Humieres <dominiq at lps dot ens.fr> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|WAITING |NEW
CC| |jakub at gcc dot gnu.org
--- Comment #7 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
% cat pr62007_db.f
program hello
integer :: i, x
C$omp parallel do simd private(x) default(none)
do i=1,10
x=x+1
end do
end program hello
triggers the same error.
^ permalink raw reply [flat|nested] 10+ messages in thread
* [Bug fortran/62007] default(none) conflicts with iteration variable in openmp parallel loop simd
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
` (7 preceding siblings ...)
2014-12-06 22:13 ` dominiq at lps dot ens.fr
@ 2014-12-08 13:32 ` jakub at gcc dot gnu.org
8 siblings, 0 replies; 10+ messages in thread
From: jakub at gcc dot gnu.org @ 2014-12-08 13:32 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62007
--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I think the message is correct, or at least the case is highly unclear, the
OpenMP 4.0 standard has various unclear corner cases.
The thing is, the loop iterator is predetermined linear, and the linear clause
makes implicit reference to the outer var, but as linear clause is not allowed
on the parallel, it can't be specified there and with default(none) you are
requesting that everything is specified there explicitly.
I can only suggest using non-combined construct in that case if you for some
reason need default(none), say:
!$omp parallel private (i_ct) default(none)
!$omp do simd
do i_ct = 1, 10
...
end do
!$omp end parallel
or so.
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2014-12-08 13:32 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-08-04 10:48 [Bug fortran/62007] New: default(none) conflicts with iteration variable in openmp parallel loop simd alexander.grund at mailbox dot tu-dresden.de
2014-08-04 10:48 ` [Bug fortran/62007] " alexander.grund at mailbox dot tu-dresden.de
2014-08-04 10:53 ` dominiq at lps dot ens.fr
2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
2014-08-04 11:02 ` alexander.grund at mailbox dot tu-dresden.de
2014-08-04 11:04 ` alexander.grund at mailbox dot tu-dresden.de
2014-08-04 12:29 ` anlauf at gmx dot de
2014-08-04 12:36 ` alexander.grund at mailbox dot tu-dresden.de
2014-12-06 22:13 ` dominiq at lps dot ens.fr
2014-12-08 13:32 ` jakub at gcc dot gnu.org
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).