April 2015 - discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] configuration issues: complex.h and CROSS_F77_SIZEOF_INTEGER
by Junchao Zhang 09 Apr '15

09 Apr '15

Lou, I looked at your config.log. You passed CC=/opt/solarisstudio12.4/bin/CC to configure, which is wrong. You should use CC=/opt/solarisstudio12.4/bin/cc. If you want C++, just use CXX=... --Junchao Zhang On Thu, Apr 9, 2015 at 9:12 AM, Junchao Zhang <jczhang(a)mcs.anl.gov> wrote: > I happened to build MPICH with SolarisStudio12.4 for Linux a couple of > days ago. > Could you send config.log of your build? > > --Junchao Zhang > > On Wed, Apr 8, 2015 at 10:51 PM, Lou Cetinich <loucetinich(a)gmail.com> > wrote: > >> I *only* need Fortran. >> >> (In fact, apart from exhaustive testing, and the endpoint target of a >> production build), I’d otherwise be happy to disable *everything else*… >> (!) >> >> Lou >> >> >> On Apr 8, 2015, at 11:47 PM, Thakur, Rajeev <thakur(a)mcs.anl.gov> wrote: >> >> If you don't need to use Fortran, you can configure with >> --disable-fortran. >> >> Rajeev >> >> On Apr 8, 2015, at 10:36 PM, Lou Picciano <loupicciano(a)comcast.net> >> wrote: >> >> Attempting to configure mpich-3.2b1 for the SolarisStudio12.4 suite... >> >> Issues around complex.h (Yes, I'll have to look for prerequisites) >> >> and the message about FORTRAN integer length - which seems to have been >> dealt with a couple of years ago, and only in the ifort context? >> >> CROSS_F77_SIZEOF_INTEGER seems to be ignored; I've tried all of 1,2,4,8... >> >> Many thanks! >> >> checking complex.h presence... yes >> configure: WARNING: complex.h: present but cannot be compiled >> configure: WARNING: complex.h: check for missing prerequisite headers? >> configure: WARNING: complex.h: see the Autoconf documentation >> configure: WARNING: complex.h: section "Present But Cannot Be >> Compiled" >> configure: WARNING: complex.h: proceeding with the compiler's result >> configure: WARNING: ## -------------------------------- ## >> configure: WARNING: ## Report this to discuss(a)mpich.org ## >> configure: WARNING: ## -------------------------------- ## >> checking for complex.h... no >> checking size of _Bool... 0 >> checking size of float _Complex... 0 >> checking size of double _Complex... 0 >> checking size of long double _Complex... 0 >> checking for _Bool... no >> checking for float _Complex... no >> checking for double _Complex... no >> checking for long double _Complex... no >> checking for size of Fortran type integer... 0 >> checking for size of Fortran type real... 0 >> checking for size of Fortran type double precision... 0 >> checking whether integer*1 is supported... yes >> checking whether integer*2 is supported... yes >> checking whether integer*4 is supported... yes >> checking whether integer*8 is supported... yes >> checking whether integer*16 is supported... no >> checking whether real*4 is supported... yes >> checking whether real*8 is supported... yes >> checking whether real*16 is supported... yes >> configure: error: Unable to configure with Fortran support because >> configure could not determine the size of a Fortran INTEGER. Consider >> setting CROSS_F77_SIZEOF_INTEGER to the length in bytes of a Fortran INTEGER >> _______________________________________________ >> discuss mailing list discuss(a)mpich.org >> To manage subscription options or unsubscribe: >> https://lists.mpich.org/mailman/listinfo/discuss >> >> >> _______________________________________________ >> discuss mailing list discuss(a)mpich.org >> To manage subscription options or unsubscribe: >> https://lists.mpich.org/mailman/listinfo/discuss >> >> >> >> _______________________________________________ >> discuss mailing list discuss(a)mpich.org >> To manage subscription options or unsubscribe: >> https://lists.mpich.org/mailman/listinfo/discuss >> > > _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] return() or exit() from int main() (NOT MPICH SPECIFIC)
by Jeff Hammond 09 Apr '15

09 Apr '15

For what it's worth, I've built MPICH 3 from source on SGI ICE and it works pretty well. Great for RMA. Collectives have issues for obvious reasons. Jeff On Thursday, April 9, 2015, Michael L. Stokes <Michael.Stokes(a)uah.edu> wrote: > Jeff, > > It is true that "mpirun" is not part of the MPI standard, but that > "really" had nothing to do with the question regarding the exit status. > > For anyone interested, I found what I was looking for in the MPI 3.0 > report. On p. 362 the MPI_Abort() call is discussed where it is stated > > ... "This function does not require that the invoking environment take any > action with the error code. However, a Unix or POSIX environment should > handle this as a return errorcode from the main program." > > It goes on say ... > > Advice to users. Whether the errorcode is returned from the executable or > from the MPI process startup mechanism (e.g., mpiexec), is an aspect of > quality of the MPI library but not mandatory. (End of advice to users.) > > Advice to implementors. Where possible, a high-quality implementation will > try to return the errorcode from the MPI process startup mechanism (e.g. > mpiexec or singleton init). (End of advice to implementors.) > > So the MPI standard is clear that returning the errorcode is not required. > However not providing it in a "high-quality implementation" for a POSIX > environment (Linux is considered POSIX.1 compliant) should be expected. So > I guess I'll have to find out if SGI considers MPT a high-quality > implementation. > > I think this also raises another point. While the MPI spec seems to > distance itself (for portability reasons) how an MPI process starts, it > does seem to care, at least partially, how it ends (see arguments to > MPI_Abort()). > > I recommend that the spec include some way that the exit status be made > available to the user environment regardless of whether MPI fails or not. > Maybe the standard should have an MPI_Exit() call ??? or maybe a argument > to MPI_Finalize() ??? > > > > On 04/09/2015 12:26 PM, Jeff Hammond wrote: > >> mpirun is not part of the MPI standard. This is something that you'll >> have to discuss with your MPI implementer at SGI. >> >> Best, >> >> Jeff >> >> On Thu, Apr 9, 2015 at 9:36 AM, Michael L. Stokes >> <Michael.Stokes(a)uah.edu> wrote: >> >>> This question is not MPICH specific, but I'm sure the expertise is here >>> to >>> answer this question. >>> >>> While running tests on spirit.afrl.hpc.mil (SGI ICE X) using the >>> MPT/2.11 >>> stack, I noticed that mpirun returns 0 to the shell regardless of the >>> exit >>> value ( <stdlib.h> exit(int) ), or the return value (return(int)) from >>> the >>> main. >>> >>> Would this behavior be regarded as an error? What are the issues? >>> >>> --Mike >>> _______________________________________________ >>> discuss mailing list discuss(a)mpich.org >>> To manage subscription options or unsubscribe: >>> https://lists.mpich.org/mailman/listinfo/discuss >>> >> >> >> > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/ _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] return() or exit() from int main() (NOT MPICH SPECIFIC)
by Michael L. Stokes 09 Apr '15

09 Apr '15

Jeff, It is true that "mpirun" is not part of the MPI standard, but that "really" had nothing to do with the question regarding the exit status. For anyone interested, I found what I was looking for in the MPI 3.0 report. On p. 362 the MPI_Abort() call is discussed where it is stated ... "This function does not require that the invoking environment take any action with the error code. However, a Unix or POSIX environment should handle this as a return errorcode from the main program." It goes on say ... Advice to users. Whether the errorcode is returned from the executable or from the MPI process startup mechanism (e.g., mpiexec), is an aspect of quality of the MPI library but not mandatory. (End of advice to users.) Advice to implementors. Where possible, a high-quality implementation will try to return the errorcode from the MPI process startup mechanism (e.g. mpiexec or singleton init). (End of advice to implementors.) So the MPI standard is clear that returning the errorcode is not required. However not providing it in a "high-quality implementation" for a POSIX environment (Linux is considered POSIX.1 compliant) should be expected. So I guess I'll have to find out if SGI considers MPT a high-quality implementation. I think this also raises another point. While the MPI spec seems to distance itself (for portability reasons) how an MPI process starts, it does seem to care, at least partially, how it ends (see arguments to MPI_Abort()). I recommend that the spec include some way that the exit status be made available to the user environment regardless of whether MPI fails or not. Maybe the standard should have an MPI_Exit() call ??? or maybe a argument to MPI_Finalize() ??? On 04/09/2015 12:26 PM, Jeff Hammond wrote: > mpirun is not part of the MPI standard. This is something that you'll > have to discuss with your MPI implementer at SGI. > > Best, > > Jeff > > On Thu, Apr 9, 2015 at 9:36 AM, Michael L. Stokes > <Michael.Stokes(a)uah.edu> wrote: >> This question is not MPICH specific, but I'm sure the expertise is here to >> answer this question. >> >> While running tests on spirit.afrl.hpc.mil (SGI ICE X) using the MPT/2.11 >> stack, I noticed that mpirun returns 0 to the shell regardless of the exit >> value ( <stdlib.h> exit(int) ), or the return value (return(int)) from the >> main. >> >> Would this behavior be regarded as an error? What are the issues? >> >> --Mike >> _______________________________________________ >> discuss mailing list discuss(a)mpich.org >> To manage subscription options or unsubscribe: >> https://lists.mpich.org/mailman/listinfo/discuss > > _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] return() or exit() from int main() (NOT MPICH SPECIFIC)
by Jeff Hammond 09 Apr '15

09 Apr '15

mpirun is not part of the MPI standard. This is something that you'll have to discuss with your MPI implementer at SGI. Best, Jeff On Thu, Apr 9, 2015 at 9:36 AM, Michael L. Stokes <Michael.Stokes(a)uah.edu> wrote: > This question is not MPICH specific, but I'm sure the expertise is here to > answer this question. > > While running tests on spirit.afrl.hpc.mil (SGI ICE X) using the MPT/2.11 > stack, I noticed that mpirun returns 0 to the shell regardless of the exit > value ( <stdlib.h> exit(int) ), or the return value (return(int)) from the > main. > > Would this behavior be regarded as an error? What are the issues? > > --Mike > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/ _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

return() or exit() from int main() (NOT MPICH SPECIFIC)
by Michael L. Stokes 09 Apr '15

09 Apr '15

This question is not MPICH specific, but I'm sure the expertise is here to answer this question. While running tests on spirit.afrl.hpc.mil (SGI ICE X) using the MPT/2.11 stack, I noticed that mpirun returns 0 to the shell regardless of the exit value ( <stdlib.h> exit(int) ), or the return value (return(int)) from the main. Would this behavior be regarded as an error? What are the issues? --Mike _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] Trouble building with PGI and IB
by Balaji, Pavan 09 Apr '15

09 Apr '15

We haven't tried building the ib netmod with PGI. We'll look into it. Thanks for reporting. In the meanwhile, did you try the mxm netmod? You can use it for Mellanox IB. -- Pavan > On Apr 8, 2015, at 5:34 PM, Martin Cuma <martin.cuma(a)utah.edu> wrote: > > Hello, > > I am trying to build MPICH2 with PGI compilers (15.1) and -with-device=ch3:nemesis:ib > and I am getting errors during the compilation. > > Here's how I configure MPICH: > setenv CC pgcc > setenv CXX pgCC > setenv FC pgf90 > setenv F77 pgf90 > setenv FC pgf90 > setenv FFLAGS "-tp=nehalem-64" > setenv FCFLAGS "-tp=nehalem-64" > setenv CFLAGS "-tp=nehalem-64" > setenv CXXFLAGS "-tp=nehalem-64" > setenv LDFLAGS "-tp=nehalem-64" > > ../../../srcdir/mpich/3.1.4/configure --prefix=/uufs/chpc.utah.edu/sys/installdir/mpich/3.1.4p --enable-romio --with-file-system=nfs+ufs --with-mpe -with-device=ch3:nemesis:ib --enable-fast=all --enable-threads=runtime > > There are multiple errors like: > PGC-S-0060-immed_len is not a member of this struct or union (../../../srcdir/mpich/3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c: 1847) > > Before I spam the list with more details, I'd like to first ask if anyone has experience with building MPICH with PGI and IB device, or if I am doing something wrong with the configuration. > > Note that MPICH builds fine with PGI and tcp and also the IB builds fine with our other supported compilers (Intel and GNU). > > Thanks, > MC > > -- > Martin Cuma > Center for High Performance Computing > Department of Geology and Geophysics > University of Utah > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Pavan Balaji ✉️ http://www.mcs.anl.gov/~balaji _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] configuration issues: complex.h and CROSS_F77_SIZEOF_INTEGER
by Junchao Zhang 09 Apr '15

09 Apr '15

I happened to build MPICH with SolarisStudio12.4 for Linux a couple of days ago. Could you send config.log of your build? --Junchao Zhang On Wed, Apr 8, 2015 at 10:51 PM, Lou Cetinich <loucetinich(a)gmail.com> wrote: > I *only* need Fortran. > > (In fact, apart from exhaustive testing, and the endpoint target of a > production build), I’d otherwise be happy to disable *everything else*… > (!) > > Lou > > > On Apr 8, 2015, at 11:47 PM, Thakur, Rajeev <thakur(a)mcs.anl.gov> wrote: > > If you don't need to use Fortran, you can configure with --disable-fortran. > > Rajeev > > On Apr 8, 2015, at 10:36 PM, Lou Picciano <loupicciano(a)comcast.net> wrote: > > Attempting to configure mpich-3.2b1 for the SolarisStudio12.4 suite... > > Issues around complex.h (Yes, I'll have to look for prerequisites) > > and the message about FORTRAN integer length - which seems to have been > dealt with a couple of years ago, and only in the ifort context? > > CROSS_F77_SIZEOF_INTEGER seems to be ignored; I've tried all of 1,2,4,8... > > Many thanks! > > checking complex.h presence... yes > configure: WARNING: complex.h: present but cannot be compiled > configure: WARNING: complex.h: check for missing prerequisite headers? > configure: WARNING: complex.h: see the Autoconf documentation > configure: WARNING: complex.h: section "Present But Cannot Be Compiled" > configure: WARNING: complex.h: proceeding with the compiler's result > configure: WARNING: ## -------------------------------- ## > configure: WARNING: ## Report this to discuss(a)mpich.org ## > configure: WARNING: ## -------------------------------- ## > checking for complex.h... no > checking size of _Bool... 0 > checking size of float _Complex... 0 > checking size of double _Complex... 0 > checking size of long double _Complex... 0 > checking for _Bool... no > checking for float _Complex... no > checking for double _Complex... no > checking for long double _Complex... no > checking for size of Fortran type integer... 0 > checking for size of Fortran type real... 0 > checking for size of Fortran type double precision... 0 > checking whether integer*1 is supported... yes > checking whether integer*2 is supported... yes > checking whether integer*4 is supported... yes > checking whether integer*8 is supported... yes > checking whether integer*16 is supported... no > checking whether real*4 is supported... yes > checking whether real*8 is supported... yes > checking whether real*16 is supported... yes > configure: error: Unable to configure with Fortran support because > configure could not determine the size of a Fortran INTEGER. Consider > setting CROSS_F77_SIZEOF_INTEGER to the length in bytes of a Fortran INTEGER > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] cross compiling using mpich
by Rob Latham 09 Apr '15

09 Apr '15

On 04/09/2015 03:33 AM, mapalo92 . wrote: > Is possible to use mpich in order to cross compile between windows and > linux? I tried to built mpich 's libraries using mingw-w64 but nothing > done. I got many many errors. Any feedback will be helpful. which is your target and which is your host? If you are building a linux target on a windows host, you might be better off setting up a virtual machine and doing your cross compilation there. In general, cross compilation works with mpich: we do it regularly for Blue Gene. ==rob -- Rob Latham Mathematics and Computer Science Division Argonne National Lab, IL USA _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] cross compiling using mpich
by Seo, Sangmin 09 Apr '15

09 Apr '15

Hello, We haven't supported Windows since MPICH version 1.4.1p1. You can find out more at the FAQ page: http://wiki.mpich.org/mpich/index.php/Frequently_Asked_Questions#Q:_Why_can… Thanks, Sangmin > On Apr 9, 2015, at 3:33 AM, mapalo92 . <mapalo2008(a)gmail.com> wrote: > > Is possible to use mpich in order to cross compile between windows and linux? I tried to built mpich 's libraries using mingw-w64 but nothing done. I got many many errors. Any feedback will be helpful. > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] Trouble building with PGI and IB
by Jeff Hammond 09 Apr '15

09 Apr '15

First, try MPICH 3.1+ not MPICH2, which is very old at this point. Second, if PGI's C compiler chokes on MPICH, use GCC's instead. And you can still use PGI's C++ and Fortran compilers to build those language bindings. As C is designed for ABI interoperability, unlike C++ or Fortran, this is perfectly fine. So, in the following, just change to CC=gcc and you should be fine. GCC won't understand "-tp=nehalem-64" so you may need to remove that from CFLAGS and LDFLAGS, but you might as well remove it completely, because it will have no effect on the C++ or Fortran bindings. Jeff On Wed, Apr 8, 2015 at 3:34 PM, Martin Cuma <martin.cuma(a)utah.edu> wrote: > Hello, > > I am trying to build MPICH2 with PGI compilers (15.1) and > -with-device=ch3:nemesis:ib > and I am getting errors during the compilation. > > Here's how I configure MPICH: > setenv CC pgcc > setenv CXX pgCC > setenv FC pgf90 > setenv F77 pgf90 > setenv FC pgf90 > setenv FFLAGS "-tp=nehalem-64" > setenv FCFLAGS "-tp=nehalem-64" > setenv CFLAGS "-tp=nehalem-64" > setenv CXXFLAGS "-tp=nehalem-64" > setenv LDFLAGS "-tp=nehalem-64" > > ../../../srcdir/mpich/3.1.4/configure > --prefix=/uufs/chpc.utah.edu/sys/installdir/mpich/3.1.4p --enable-romio > --with-file-system=nfs+ufs --with-mpe -with-device=ch3:nemesis:ib > --enable-fast=all --enable-threads=runtime > > There are multiple errors like: > PGC-S-0060-immed_len is not a member of this struct or union > (../../../srcdir/mpich/3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c: > 1847) > > Before I spam the list with more details, I'd like to first ask if anyone > has experience with building MPICH with PGI and IB device, or if I am doing > something wrong with the configuration. > > Note that MPICH builds fine with PGI and tcp and also the IB builds fine > with our other supported compilers (Intel and GNU). > > Thanks, > MC > > -- > Martin Cuma > Center for High Performance Computing > Department of Geology and Geophysics > University of Utah > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/ _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0