February 2018 - mpiwg-fortran - mailman3.cels.anl.gov

Re: [MPIWG Fortran] [Mpi-comments] MPI_Comm_spawn_multiple: parameter description and Fortran bindings
by Bill Long 09 Feb '18

09 Feb '18

My reading of example 10.2 (in the 3.0 standard) is that the command array has 2 elements, (same as the value of count), but that an extra blank element is used to signal the end of the argument lists for each command in the array array_of_argv. That seems reasonable since the number of arguments would, in general, be different for each command. Here is how you do it in Fortran: CHARACTER*25 commands(2), array_of_argv(2, 3) commands(1) = ’ ocean ’ array_of_argv(1, 1) = ’ -gridfile ’ array_of_argv(1, 2) = ’ ocean1.grd’ array_of_argv(1, 3) = ’ ’ commands(2) = ’ atmos ’ array_of_argv(2, 1) = ’ atmos.grd ’ array_of_argv(2, 2) = ’ ’ call MPI_COMM_SPAWN_MULTIPLE(2, commands, array_of_argv, ...) Self-documentation of the declaration would be clearer if the command argument were declared command(count) instead of command(*). I suspect that the (*) declaration of the command dummy argument was done that way out of habit. Cheers, Bill > On Feb 9, 2018, at 12:26 AM, Gilles Gouaillardet <gilles.gouaillardet(a)gmail.com> wrote: > > Bill, > >>>> That being said, the standard does not state that, and it is unclear whether 'array_of_commands' should be >>>> >>>> NULL-terminated (so to speak) or not. >>>> >> >> Fortran characters are NOT null terminated. > > > I was referring the array and not the characters, so let me clarify this : > > > In example 10.2, the array of commands is > > CHARACTER*25 commands(2), > commands(1) = ' ocean ' > commands(2) = ' atmos ' > > By reading the standard literally, it was unclear to me whether this > should have been > > CHARACTER*25 commands(3), > commands(1) = ' ocean ' > commands(2) = ' atmos ' > commands(3) = ' ' > > > FWIW, Open MPI currently expects the latter form (see > https://github.com/open-mpi/ompi/issues/4788) > I initially thought it was a simple bug, but on second thought, I am > still on the fence whether this should be > regarded as a simple bug or a correct/literal interpretation of the standard. > > > Cheers, > > Gilles > > On Fri, Feb 9, 2018 at 7:07 AM, Bill Long <longb(a)cray.com> wrote: >> >>> On Feb 8, 2018, at 3:44 PM, Jeff Squyres (jsquyres) <jsquyres(a)cisco.com> wrote: >>> >>>> >>>> And the Fortran binding is >>>> >>>> MPI_COMM_SPAWN_MULTIPLE(COUNT, ARRAY_OF_COMMANDS, ...) >>>> INTEGER COUNT >>>> CHARACTER*(*) ARRAY_OF_COMMANDS(*) >>>> >> >> Note that the character*(*) syntax is archaic (as of 25+ years ago). The normal way of writing this is character(*). >> >>>> >>>> Intuitively, it seems obvious that 'array_of_commands' is an array of 'count' commands. >> >> That is what I would assume. The user would be responsible for declaring the array elements large enough to hold the longest command. (All the elements of the array have the same length). >> >>>> >>>> That being said, the standard does not state that, and it is unclear whether 'array_of_commands' should be >>>> >>>> NULL-terminated (so to speak) or not. >>>> >> >> Fortran characters are NOT null terminated. >> >>>> >>>> Back to the Fortran binding, the following prototype, which is somehow consistent with 'array_of_argv', >>>> >>>> could (help to) clear the confusion. The same comments also apply to the Fortran 2008 binding. >>>> >>>> >>>> CHARACTER*(*) ARRAY_OF_COMMANDS(COUNT) >> >> This is also fine, and clearer in terms of the expected minimum size of ARRAY_OF_COMMANDS. >> >> Cheers, >> Bill >> >> >>>> >>>> >> >> Bill Long longb(a)cray.com >> Principal Engineer, Fortran Technical Support & voice: 651-605-9024 >> Bioinformatics Software Development fax: 651-605-9143 >> Cray Inc./ 2131 Lindau Lane/ Suite 1000/ Bloomington, MN 55425 >> >> > _______________________________________________ > mpiwg-fortran mailing list > mpiwg-fortran(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpiwg-fortran Bill Long longb(a)cray.com Principal Engineer, Fortran Technical Support & voice: 651-605-9024 Bioinformatics Software Development fax: 651-605-9143 Cray Inc./ 2131 Lindau Lane/ Suite 1000/ Bloomington, MN 55425

1 0

Re: [MPIWG Fortran] [Mpi-comments] MPI_Comm_spawn_multiple: parameter description and Fortran bindings
by Gilles Gouaillardet 09 Feb '18

09 Feb '18

Bill, >>> That being said, the standard does not state that, and it is unclear whether 'array_of_commands' should be >>> >>> NULL-terminated (so to speak) or not. >>> > > Fortran characters are NOT null terminated. I was referring the array and not the characters, so let me clarify this : In example 10.2, the array of commands is CHARACTER*25 commands(2), commands(1) = ' ocean ' commands(2) = ' atmos ' By reading the standard literally, it was unclear to me whether this should have been CHARACTER*25 commands(3), commands(1) = ' ocean ' commands(2) = ' atmos ' commands(3) = ' ' FWIW, Open MPI currently expects the latter form (see https://github.com/open-mpi/ompi/issues/4788) I initially thought it was a simple bug, but on second thought, I am still on the fence whether this should be regarded as a simple bug or a correct/literal interpretation of the standard. Cheers, Gilles On Fri, Feb 9, 2018 at 7:07 AM, Bill Long <longb(a)cray.com> wrote: > >> On Feb 8, 2018, at 3:44 PM, Jeff Squyres (jsquyres) <jsquyres(a)cisco.com> wrote: >> >>> >>> And the Fortran binding is >>> >>> MPI_COMM_SPAWN_MULTIPLE(COUNT, ARRAY_OF_COMMANDS, ...) >>> INTEGER COUNT >>> CHARACTER*(*) ARRAY_OF_COMMANDS(*) >>> > > Note that the character*(*) syntax is archaic (as of 25+ years ago). The normal way of writing this is character(*). > >>> >>> Intuitively, it seems obvious that 'array_of_commands' is an array of 'count' commands. > > That is what I would assume. The user would be responsible for declaring the array elements large enough to hold the longest command. (All the elements of the array have the same length). > >>> >>> That being said, the standard does not state that, and it is unclear whether 'array_of_commands' should be >>> >>> NULL-terminated (so to speak) or not. >>> > > Fortran characters are NOT null terminated. > >>> >>> Back to the Fortran binding, the following prototype, which is somehow consistent with 'array_of_argv', >>> >>> could (help to) clear the confusion. The same comments also apply to the Fortran 2008 binding. >>> >>> >>> CHARACTER*(*) ARRAY_OF_COMMANDS(COUNT) > > This is also fine, and clearer in terms of the expected minimum size of ARRAY_OF_COMMANDS. > > Cheers, > Bill > > >>> >>> > > Bill Long longb(a)cray.com > Principal Engineer, Fortran Technical Support & voice: 651-605-9024 > Bioinformatics Software Development fax: 651-605-9143 > Cray Inc./ 2131 Lindau Lane/ Suite 1000/ Bloomington, MN 55425 > >

1 0

Re: [MPIWG Fortran] [Mpi-comments] MPI_Comm_spawn_multiple: parameter description and Fortran bindings
by Bill Long 08 Feb '18

08 Feb '18

> On Feb 8, 2018, at 3:44 PM, Jeff Squyres (jsquyres) <jsquyres(a)cisco.com> wrote: > >> >> And the Fortran binding is >> >> MPI_COMM_SPAWN_MULTIPLE(COUNT, ARRAY_OF_COMMANDS, ...) >> INTEGER COUNT >> CHARACTER*(*) ARRAY_OF_COMMANDS(*) >> Note that the character*(*) syntax is archaic (as of 25+ years ago). The normal way of writing this is character(*). >> >> Intuitively, it seems obvious that 'array_of_commands' is an array of 'count' commands. That is what I would assume. The user would be responsible for declaring the array elements large enough to hold the longest command. (All the elements of the array have the same length). >> >> That being said, the standard does not state that, and it is unclear whether 'array_of_commands' should be >> >> NULL-terminated (so to speak) or not. >> Fortran characters are NOT null terminated. >> >> Back to the Fortran binding, the following prototype, which is somehow consistent with 'array_of_argv', >> >> could (help to) clear the confusion. The same comments also apply to the Fortran 2008 binding. >> >> >> CHARACTER*(*) ARRAY_OF_COMMANDS(COUNT) This is also fine, and clearer in terms of the expected minimum size of ARRAY_OF_COMMANDS. Cheers, Bill >> >> Bill Long longb(a)cray.com Principal Engineer, Fortran Technical Support & voice: 651-605-9024 Bioinformatics Software Development fax: 651-605-9143 Cray Inc./ 2131 Lindau Lane/ Suite 1000/ Bloomington, MN 55425

1 0

Re: [MPIWG Fortran] Fwd: [Mpi-comments] MPI_Comm_spawn_multiple: parameter description and Fortran bindings
by Jeff Hammond 08 Feb '18

08 Feb '18

I'd rather deprecate MPI_Comm_spawn_multiple than try to wrap my head around this one. Jeff On Thu, Feb 8, 2018 at 1:44 PM, Jeff Squyres (jsquyres) <jsquyres(a)cisco.com> wrote: > > Forwarding this from the MPI-comments list to the Fortran working group... > > > Begin forwarded message: > > From: Gilles Gouaillardet <gilles(a)rist.or.jp> > Subject: [Mpi-comments] MPI_Comm_spawn_multiple: parameter description and Fortran bindings > Date: February 6, 2018 at 3:04:00 AM EST > To: <mpi-comments(a)lists.mpi-forum.org> > Reply-To: Comments on the MPI Standard <mpi-comments(a)lists.mpi-forum.org> > > Dear MPI Forum, > > > The description of the 'array_of_commands` parameter passed to 'MPI_Comm_spawn_multiple()' is not crystal clear IMHO. > > The standard simply states > > MPI_COMM_SPAWN_MULTIPLE(count, array_of_commands, ...) > IN count number of commands (positive integer, significant to MPI > only at root | see advice to users) > IN array_of_commands programs to be executed (array of strings, > significant only at root) > > Advice to users. The argument count is interpreted by MPI only at > the root, as is > array_of_argv. Since the leading dimension of array_of_argv is > count, a non-positive > value of count at a non-root node could theoretically cause a > runtime bounds check > error, even though array_of_argv should be ignored by the > subroutine. If this happens, > you should explicitly supply a reasonable value of count on the > non-root nodes. (End > of advice to users.) > > > And the Fortran binding is > > MPI_COMM_SPAWN_MULTIPLE(COUNT, ARRAY_OF_COMMANDS, ...) > INTEGER COUNT > CHARACTER*(*) ARRAY_OF_COMMANDS(*) > > > Intuitively, it seems obvious that 'array_of_commands' is an array of 'count' commands. > > That being said, the standard does not state that, and it is unclear whether 'array_of_commands' should be > > NULL-terminated (so to speak) or not. > > > Back to the Fortran binding, the following prototype, which is somehow consistent with 'array_of_argv', > > could (help to) clear the confusion. The same comments also apply to the Fortran 2008 binding. > > > CHARACTER*(*) ARRAY_OF_COMMANDS(COUNT) > > > > Gilles > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpi-comments > > > > -- > Jeff Squyres > jsquyres(a)cisco.com > > > > > _______________________________________________ > mpiwg-fortran mailing list > mpiwg-fortran(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpiwg-fortran -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/

1 0

Fwd: [Mpi-comments] MPI_Comm_spawn_multiple: parameter description and Fortran bindings
by Jeff Squyres (jsquyres) 08 Feb '18

08 Feb '18

Forwarding this from the MPI-comments list to the Fortran working group... Begin forwarded message: From: Gilles Gouaillardet <gilles(a)rist.or.jp<mailto:[email protected]>> Subject: [Mpi-comments] MPI_Comm_spawn_multiple: parameter description and Fortran bindings Date: February 6, 2018 at 3:04:00 AM EST To: <mpi-comments(a)lists.mpi-forum.org<mailto:[email protected]>> Reply-To: Comments on the MPI Standard <mpi-comments(a)lists.mpi-forum.org<mailto:[email protected]>> Dear MPI Forum, The description of the 'array_of_commands` parameter passed to 'MPI_Comm_spawn_multiple()' is not crystal clear IMHO. The standard simply states MPI_COMM_SPAWN_MULTIPLE(count, array_of_commands, ...) IN count number of commands (positive integer, significant to MPI only at root | see advice to users) IN array_of_commands programs to be executed (array of strings, significant only at root) Advice to users. The argument count is interpreted by MPI only at the root, as is array_of_argv. Since the leading dimension of array_of_argv is count, a non-positive value of count at a non-root node could theoretically cause a runtime bounds check error, even though array_of_argv should be ignored by the subroutine. If this happens, you should explicitly supply a reasonable value of count on the non-root nodes. (End of advice to users.) And the Fortran binding is MPI_COMM_SPAWN_MULTIPLE(COUNT, ARRAY_OF_COMMANDS, ...) INTEGER COUNT CHARACTER*(*) ARRAY_OF_COMMANDS(*) Intuitively, it seems obvious that 'array_of_commands' is an array of 'count' commands. That being said, the standard does not state that, and it is unclear whether 'array_of_commands' should be NULL-terminated (so to speak) or not. Back to the Fortran binding, the following prototype, which is somehow consistent with 'array_of_argv', could (help to) clear the confusion. The same comments also apply to the Fortran 2008 binding. CHARACTER*(*) ARRAY_OF_COMMANDS(COUNT) Gilles _______________________________________________ mpi-comments mailing list mpi-comments(a)lists.mpi-forum.org<mailto:[email protected]> https://lists.mpi-forum.org/mailman/listinfo/mpi-comments -- Jeff Squyres jsquyres(a)cisco.com<mailto:[email protected]>

1 0