current status of openmp

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

current status of openmp

Jack Howarth
      Looking at the portions of the OpenMP3.1_Validation which
currently fail on x86_64-apple-darwin14 at r236714 with
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20150504/128665.html
applied, I see...

#Tested Directive         t ct ot oct

omp_threadprivate         100 100 - -

omp_task_untied           0 - 0 -

omp_taskyield             100 100 10 -

omp_task_final            0 - 0 -

Perhaps the openmp maintainers could expound on which of these will be
fixed for the 3.7 release and which are optional features which won't
be implemented.
             Jack

ps
    Certain test regress in their verification level if 'make ctest'
is repeated enough times. Is this a defect in the OpenMP3.1_Validation
test suite or a fragility in the current openmp support for those
features?

omp_flush                 100 95 100 100

omp_for_lastprivate        100 100 100 80

omp_flush                 100 100 100 95

omp_parallel_sections_reduction 100 100 100 95
_______________________________________________
cfe-dev mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
Reply | Threaded
Open this post in threaded view
|

Re: [Openmp-dev] current status of openmp

Alexey Bataev
Jack, untied tasks and final task will be supportd soon.
As to tests from test suite, yes, they are fragile and sometimes just
unable to verify constructs completely. Also, it seems to me there is a
bug in tests for threadprivate construct, that's why there

- - in orphaned tests

Best regards,
Alexey Bataev
=============
Software Engineer
Intel Compiler Team

07.05.2015 16:02, Jack Howarth пишет:

>        Looking at the portions of the OpenMP3.1_Validation which
> currently fail on x86_64-apple-darwin14 at r236714 with
> http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20150504/128665.html
> applied, I see...
>
> #Tested Directive         t ct ot oct
>
> omp_threadprivate         100 100 - -
>
> omp_task_untied           0 - 0 -
>
> omp_taskyield             100 100 10 -
>
> omp_task_final            0 - 0 -
>
> Perhaps the openmp maintainers could expound on which of these will be
> fixed for the 3.7 release and which are optional features which won't
> be implemented.
>               Jack
>
> ps
>      Certain test regress in their verification level if 'make ctest'
> is repeated enough times. Is this a defect in the OpenMP3.1_Validation
> test suite or a fragility in the current openmp support for those
> features?
>
> omp_flush                 100 95 100 100
>
> omp_for_lastprivate        100 100 100 80
>
> omp_flush                 100 100 100 95
>
> omp_parallel_sections_reduction 100 100 100 95
> _______________________________________________
> Openmp-dev mailing list
> [hidden email]
> http://lists.cs.uiuc.edu/mailman/listinfo/openmp-dev


_______________________________________________
cfe-dev mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
--
Best regards,
Alexey Bataev
=============
Software Engineer
Intel Compiler Team
Intel Corp.
Reply | Threaded
Open this post in threaded view
|

Re: [Openmp-dev] current status of openmp

Sunita Chandrasekaran
I see, we will double check the threadprivate test. 
Thanks
Sunita

--
Sunita Chandrasekaran
University of Houston


On Thu, May 7, 2015 at 10:06 PM, Bataev, Alexey <[hidden email]> wrote:
Jack, untied tasks and final task will be supportd soon.
As to tests from test suite, yes, they are fragile and sometimes just unable to verify constructs completely. Also, it seems to me there is a bug in tests for threadprivate construct, that's why there

- - in orphaned tests

Best regards,
Alexey Bataev
=============
Software Engineer
Intel Compiler Team

07.05.2015 16:02, Jack Howarth пишет:

       Looking at the portions of the OpenMP3.1_Validation which
currently fail on x86_64-apple-darwin14 at r236714 with
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20150504/128665.html
applied, I see...

#Tested Directive         t ct ot oct

omp_threadprivate         100 100 - -

omp_task_untied           0 - 0 -

omp_taskyield             100 100 10 -

omp_task_final            0 - 0 -

Perhaps the openmp maintainers could expound on which of these will be
fixed for the 3.7 release and which are optional features which won't
be implemented.
              Jack

ps
     Certain test regress in their verification level if 'make ctest'
is repeated enough times. Is this a defect in the OpenMP3.1_Validation
test suite or a fragility in the current openmp support for those
features?

omp_flush                 100 95 100 100

omp_for_lastprivate        100 100 100 80

omp_flush                 100 100 100 95

omp_parallel_sections_reduction 100 100 100 95
_______________________________________________
Openmp-dev mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/openmp-dev


_______________________________________________
Openmp-dev mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/openmp-dev


_______________________________________________
cfe-dev mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev