October 2011 - mpi-forum - mailman3.cels.anl.gov

Re: [Mpi-forum] Oct Meeting
by Jeff Squyres 04 Oct '11

04 Oct '11

Thanks. FWIW, I registered because the web form had the correct date on the top. So I assumed that it had been reset for the Chicago meeting. On Oct 4, 2011, at 8:11 AM, Graham, Richard L. wrote: > I not sure - someone will contact those that registered before the page was ready, and let you know. > > Rich > > > > Sent with Good (www.good.com) > > > -----Original Message----- > From: Jeff Squyres [mailto:[email protected]] > Sent: Tuesday, October 04, 2011 07:56 AM Eastern Standard Time > To: Main MPI Forum mailing list > Subject: Re: [Mpi-forum] Oct Meeting > > Do you mean that we need to fill out the form again? > > I already filled it out, got the confirmation mail, and see the charge on my Amex. > > > On Oct 4, 2011, at 7:52 AM, RICHARD GRAHAM wrote: > >> The meeting registration page for the Oct 24-26 meeting is up. Note that those who registered before the page was announced will need to re-do their registration (there was an error in the cost for the meeting), but not sure how this will be handled. >> >> Please send me requests for agenda items. >> >> Thanks, >> Rich >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> hxxp://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: > hxxp://www.cisco.com/web/about/doing_business/legal/cri/ > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > hxxp://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi-forum] Oct Meeting
by Graham, Richard L. 04 Oct '11

04 Oct '11

I not sure - someone will contact those that registered before the page was ready, and let you know. Rich Sent with Good (www.good.com) -----Original Message----- From: Jeff Squyres [mailto:[email protected]] Sent: Tuesday, October 04, 2011 07:56 AM Eastern Standard Time To: Main MPI Forum mailing list Subject: Re: [Mpi-forum] Oct Meeting Do you mean that we need to fill out the form again? I already filled it out, got the confirmation mail, and see the charge on my Amex. On Oct 4, 2011, at 7:52 AM, RICHARD GRAHAM wrote: > The meeting registration page for the Oct 24-26 meeting is up. Note that those who registered before the page was announced will need to re-do their registration (there was an error in the cost for the meeting), but not sure how this will be handled. > > Please send me requests for agenda items. > > Thanks, > Rich > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > hxxp://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: hxxp://www.cisco.com/web/about/doing_business/legal/cri/ _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org hxxp://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

Re: [Mpi-forum] Oct Meeting
by Jeff Squyres 04 Oct '11

04 Oct '11

Do you mean that we need to fill out the form again? I already filled it out, got the confirmation mail, and see the charge on my Amex. On Oct 4, 2011, at 7:52 AM, RICHARD GRAHAM wrote: > The meeting registration page for the Oct 24-26 meeting is up. Note that those who registered before the page was announced will need to re-do their registration (there was an error in the cost for the meeting), but not sure how this will be handled. > > Please send me requests for agenda items. > > Thanks, > Rich > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Oct Meeting
by RICHARD GRAHAM 04 Oct '11

04 Oct '11

The meeting registration page for the Oct 24-26 meeting is up. Note that those who registered before the page was announced will need to re-do their registration (there was an error in the cost for the meeting), but not sure how this will be handled. Please send me requests for agenda items. Thanks, Rich

1 0

Re: [Mpi-forum] Fortran bindings
by Graham, Richard L. 03 Oct '11

03 Oct '11

comments in line: On Oct 3, 2011, at 1:56 PM, Rolf Rabenseifner wrote: >> - Be able to use all three versions of the Fortran bindings within a >> single translation unit. The goal being to make it possible to migrate >> incrementally to the new proposed bindings, within the applications >> current file structure. > > This is only one aspect of the proposed Fortran interface. > The major aspect is to solve the inconsistency problems between > Fortran and MPI. > And here mainly the problems with nonblocking routines. [rich] As I understand this, the interface and the Fortran/MPI interoperability are two distinct issues. That is, the API and the issues that arise from compiler optimizations are two distinct items. >> - Take advantage of the ability to have Fortran and C binding be >> exactly the same, bypassing a translation layer. > > This is not a choice because it does not solve the major problem: > The correct use of nonblocking routines. [rich] Again, if I understood this correctly, the form of the interface is a different issue than what to do to prevent the compiler optimizations that cause issues with nonblocking routines. > > Using the C interface from Fortran means that the user has to > - translate all the handles from Fortran to C and vice versa, > - pass the C_LOC(buffer) instead of buffer, > - use the TARGET attribute for the buffer > (otherwise the use of C_LOC with buffer is incorrect). [rich] Can the arguments in fortran be declares as c types, avoiding all the translation you mention ? > > The TARGET attribute does not solve the nonblocking problem > because it protects only code movement optimization > around the MPI_Wait calls, but it does not solve problems > based on automatic off-loading of data to a GPU memory > while an internal MPI progress thread is communicating > parts of the array. > Such "using only C from Fortran" may be only portable when > there is no progress thread as, e.g., in mpich2. > > The MPI Forum clearly decided ***not*** to mandate one > of these approaches (with or without a progress thread), > and therefore, a pure TARGET based solution and to > forbid progress threads is not an option compared to > the ASYNCHRONOUS buffers in the new mpi_f08 module. [rich] The forum has not decided anything at this stage, which is why I ask these questions before we go ahead and actually make decisions. > > If users want to use the C interface from Fortran, > they can do it, but they should know that they write > non-portable applications. > There are real user who do this, as Torsten reported. [rich] The question was strictly about the interface. I think that what Craig said, is that the current proposal is what Fortran users expect is what answers my question. Rich > > Best regards > Rolf > > > ----- Original Message ----- >> From: "Richard L. Graham" <rlgraham(a)ornl.gov> >> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> >> Sent: Monday, October 3, 2011 6:30:43 PM >> Subject: [Mpi-forum] Fortran bindings >> At the risk of upsetting the apple cart, I want to ask a question >> about the new proposed Fortran bindings for MPI 3.0. I rather sure >> that the question I am about to ask has been addressed in the past, >> but to be honest, I did not understand the details of the proposal >> enough to note the difference. It was not until the rather heated (and >> fruitful) discussions that occurred at the last meeting in Santorini, >> and a follow on discussion that this came up. >> >> It seems to me that two options were considered with respect to these >> bindings: >> - Be able to use all three versions of the Fortran bindings within a >> single translation unit. The goal being to make it possible to migrate >> incrementally to the new proposed bindings, within the applications >> current file structure. >> - Take advantage of the ability to have Fortran and C binding be >> exactly the same, bypassing a translation layer. >> >> First, did I describe the two possibilities correctly ? If so, why opt >> for the first over the second ? >> >> Rich >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> hxxp://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > -- > Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de > High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 > University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 > Head of Dpmt Parallel Computing . . . www . hlrs . de/people/rabenseifner > Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30) > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > hxxp://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >

1 0

Re: [Mpi-forum] Fortran bindings
by Graham, Richard L. 03 Oct '11

03 Oct '11

OK - forgot about the return parameter - so thanks a lot !!! Rich On Oct 3, 2011, at 1:27 PM, Rasmussen, Craig E wrote: > See answers to your questions below. > > -craig > > > On Oct 3, 2011, at 10:30 AM, Graham, Richard L. wrote: > >> At the risk of upsetting the apple cart, I want to ask a question about the new proposed Fortran bindings for MPI 3.0. I rather sure that the question I am about to ask has been addressed in the past, but to be honest, I did not understand the details of the proposal enough to note the difference. It was not until the rather heated (and fruitful) discussions that occurred at the last meeting in Santorini, and a follow on discussion that this came up. >> >> It seems to me that two options were considered with respect to these bindings: >> - Be able to use all three versions of the Fortran bindings within a single translation unit. The goal being to make it possible to migrate incrementally to the new proposed bindings, within the applications current file structure. > > Yes. > >> - Take advantage of the ability to have Fortran and C binding be exactly the same, bypassing a translation layer. > > No. We started down this path at the beginning. However, this would have required that the Fortran procedures be functions rather than subroutines. Early on we discussed this at length and decided that users would prefer the usage of subroutines (as is currently done in MPI). Using functions in Fortran REQUIRES setting the return value with a variable in an assignment statement. So standard practice in Fortran is to user subroutines for most procedures. > > The choice of using subroutines (a function that returns a void) means that a wrapper must be used to convert the C MPI function call. In addition, since we have to have wrappers anyway, we decided to make MPI handles be a user defined type (think C struct) containing a single integer. This removes a lot of messy conversions routines between new handles, C handles, and older MPI handles. It also will allow the older wrapper functions to be called directly for most routines so implementers won't have to maintain two sets of wrappers. The wrappers can be written in either Fortran or C. > > -craig > >> >> First, did I describe the two possibilities correctly ? If so, why opt for the first over the second ? >> > > I don't think these two options are mutually exclusive. You could have both but we decided early on against the second. > > -craig > >

1 0

Re: [Mpi-forum] Fortran bindings
by Rolf Rabenseifner 03 Oct '11

03 Oct '11

> - Be able to use all three versions of the Fortran bindings within a > single translation unit. The goal being to make it possible to migrate > incrementally to the new proposed bindings, within the applications > current file structure. This is only one aspect of the proposed Fortran interface. The major aspect is to solve the inconsistency problems between Fortran and MPI. And here mainly the problems with nonblocking routines. > - Take advantage of the ability to have Fortran and C binding be > exactly the same, bypassing a translation layer. This is not a choice because it does not solve the major problem: The correct use of nonblocking routines. Using the C interface from Fortran means that the user has to - translate all the handles from Fortran to C and vice versa, - pass the C_LOC(buffer) instead of buffer, - use the TARGET attribute for the buffer (otherwise the use of C_LOC with buffer is incorrect). The TARGET attribute does not solve the nonblocking problem because it protects only code movement optimization around the MPI_Wait calls, but it does not solve problems based on automatic off-loading of data to a GPU memory while an internal MPI progress thread is communicating parts of the array. Such "using only C from Fortran" may be only portable when there is no progress thread as, e.g., in mpich2. The MPI Forum clearly decided ***not*** to mandate one of these approaches (with or without a progress thread), and therefore, a pure TARGET based solution and to forbid progress threads is not an option compared to the ASYNCHRONOUS buffers in the new mpi_f08 module. If users want to use the C interface from Fortran, they can do it, but they should know that they write non-portable applications. There are real user who do this, as Torsten reported. Best regards Rolf ----- Original Message ----- > From: "Richard L. Graham" <rlgraham(a)ornl.gov> > To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> > Sent: Monday, October 3, 2011 6:30:43 PM > Subject: [Mpi-forum] Fortran bindings > At the risk of upsetting the apple cart, I want to ask a question > about the new proposed Fortran bindings for MPI 3.0. I rather sure > that the question I am about to ask has been addressed in the past, > but to be honest, I did not understand the details of the proposal > enough to note the difference. It was not until the rather heated (and > fruitful) discussions that occurred at the last meeting in Santorini, > and a follow on discussion that this came up. > > It seems to me that two options were considered with respect to these > bindings: > - Be able to use all three versions of the Fortran bindings within a > single translation unit. The goal being to make it possible to migrate > incrementally to the new proposed bindings, within the applications > current file structure. > - Take advantage of the ability to have Fortran and C binding be > exactly the same, bypassing a translation layer. > > First, did I describe the two possibilities correctly ? If so, why opt > for the first over the second ? > > Rich > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)

1 0

Fortran bindings
by Graham, Richard L. 03 Oct '11

03 Oct '11

At the risk of upsetting the apple cart, I want to ask a question about the new proposed Fortran bindings for MPI 3.0. I rather sure that the question I am about to ask has been addressed in the past, but to be honest, I did not understand the details of the proposal enough to note the difference. It was not until the rather heated (and fruitful) discussions that occurred at the last meeting in Santorini, and a follow on discussion that this came up. It seems to me that two options were considered with respect to these bindings: - Be able to use all three versions of the Fortran bindings within a single translation unit. The goal being to make it possible to migrate incrementally to the new proposed bindings, within the applications current file structure. - Take advantage of the ability to have Fortran and C binding be exactly the same, bypassing a translation layer. First, did I describe the two possibilities correctly ? If so, why opt for the first over the second ? Rich

1 0