https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92311
--- Comment #8 from Nichols A. Romero <naromero at anl dot gov> --- (In reply to Tobias Burnus from comment #6) > (In reply to Nichols A. Romero from comment #0) > > For the OpenACC program, it may be that this is part of the OpenACC spec > > that has not been implemented yet. > > [That's detach/attach] This OpenACC 2.6 feature is not yet available in GCC > 9 or the development version (10.0); however, it is available in the vendor > branch openacc-gcc-9 (via GCC's git repository) and there are plans to merge > it in GCC 10, cf. https://gcc.gnu.org/ml/gcc-patches/2019-10/msg00444.html > If all posted OpenACC patches get merged, expect a better/more complete > OpenACC / OpenACC 2.6 support in GCC 10. > > (In reply to Jakub Jelinek from comment #1) > > (In reply to Nichols A. Romero from comment #0) > > > You don't need a map clause here, use_device_ptr clause is enough. > > That is what OpenMP 5.0 says, but not what 4.5 said, 4.5 required at least > > one map clause, and GCC 9 implements in Fortran only OpenMP 4.0 and some > > parts of OpenMP 4.5, not OpenMP 5.0. > > GCC 9 support leaving out the map clause for C/C++; GCC 10 adds support of > more of OpenMP 5 (esp. for C/C++). However, OpenMP support in GCC's Fortran > compiler is lagging behind. Given that GCC development stage 1 ends in two > weeks, it is unlikely that GCC 10 will support leaving out 'map' in Fortran > programs. Sorry. Tobias, Thanks for your comments. Where do I find the OpenACC and OpenMP development roadmap for Fortran? Do I just search the GCC mailing list archive? I apologize for sending you and Jakub on a wild goose chase *twice* with my faulty test cases. For some context, I am writing an OpenACC -> OpenMP converter for Fortran programs as part of the DOE Exascale Computing Project. My original goal was to figure out how to translate OpenACC detach/attach to OpenMP. I was told to use use_device_ptr. But that doesn't seem to be necessary at all. It looks like OpenACC detach/attach is already handle by OpenMP's map clause (but I am still unsure as I have not been able to find someone who understands both specs. very well).
