[mvapich-discuss] MVAPICH2 build problems with PGI compiler and Slurm

[Jonathan Perkins]

Hello Martin.  This problem is coming about because of an interaction with
a libtool file present in your slurm installation.

My suggestion is to temporarily (or permanently if you always used shared
libraries) rename  the libpmi.la file in the lib or lib64 directory of your
slurm installation to something like libpmi.la.bak.

The issue comes because libtool will try to use the options present in any
.la file it finds for libraries that its linking against.  In this case the
slurm library was built with a compiler that used the -pthread option
whereas PGI doesn't support this option.

Let me know if this helps.

On Tue, Mar 31, 2015 at 11:47 AM Martin Cuma <martin.cuma at utah.edu> wrote:

> Hello,
>
> we are deploying Slurm at our center and as such I am building MVAPICH2
> with Slurm support. All goes well with GNU and Intel compilers, but, with
> PGI (15.1) during the make phase I get the following error:
>
> pgf90-Error-Unknown switch: -pthread
>
> This happens during the library collection phase using the libtool.
>
> Below is my configure setup:
> setenv CC pgcc
> setenv CXX pgCC
> setenv FC pgf90
> setenv F77 pgf90
> setenv SLURMLIB "/uufs/chpc.utah.edu/sys/pkg/slurm/std/lib"
> setenv SLURMINC "/uufs/chpc.utah.edu/sys/pkg/slurm/std/include"
> setenv FFLAGS   " -fPIC -I$SLURMINC"
> setenv FCFLAGS  " -fPIC -I$SLURMINC"
> setenv CFLAGS   " -fPIC -I$SLURMINC"
> setenv CXXFLAGS " -fPIC -I$SLURMINC"
> setenv CPPFLAGS " -fPIC -I$SLURMINC"
> setenv LDFLAGS  " -fPIC -L$SLURMLIB"
>
> ../../../srcdir/mvapich2/2.1rc2/configure
> --prefix=/uufs/chpc.utah.edu/sys/installdir/mvapich2/2.1rc2p1
> --enable-romio --with-file-system=nfs+ufs --enable-sharedlibs=gcc
> --with-ib-include=/usr/include --with-ib-libpath=/usr/lib64
> --enable-threads=runtime --enable-shared --with-rdma=gen2 --enable-cxx
> --with-pmi=pmi1 --with-pm=slurm
>
> Note that if I remove --with-pmi=pmi1 --with-pm=slurm, the build runs
> fine.
>
> Attached is the full make output of the problematic phase.
>
> This is on a RHEL 6.6 system.
>
> I appreciate if you could look at this.
>
> Thanks,
> MC
>
> --
> Martin Cuma
> Center for High Performance Computing
> Department of Geology and Geophysics
> University of Utah_______________________________________________
> mvapich-discuss mailing list
> mvapich-discuss at cse.ohio-state.edu
> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20150331/49e465b8/attachment.html>