[mvapich-discuss] Help Building MVAPICH2 with NAG 6.1: pthread fun
Thompson, Matt (GSFC-610.1)[SCIENCE SYSTEMS AND APPLICATIONS INC]
matthew.thompson at nasa.gov
Wed Jan 11 12:53:49 EST 2017
MVAPICH2 Gurus,
I'm trying to build MVAPICH2 2.2 with nagfor 6.1 and gcc 5.3 (for the
C/C++). I was able to configure with:
> ./configure --prefix=$SWDEV/MPI/mvapich2/2.2/nagfor_6.1-gcc_5.3\
--disable-wrapper-rpath --with-device=ch3:mrail --with-rdma=gen2 \
--with-pmi=pmi2 --with-pm=slurm --with-slurm=/usr/slurm \
CC=gcc CXX=g++ FC=nagfor F77=nagfor CFLAGS='-fpic -m64' \
CXXFLAGS='-fpic -m64' FCFLAGS='-pic -mismatch' FFLAGS='-pic -mismatch' \
--enable-fortran=all --enable-cxx --enable-romio --with-hwloc \
--disable-multi-aliases --enable-xrc=yes --enable-hybrid | & tee
configure.nagfor_6.1-gcc_5.3.log
and all seemed well. (Note: Many of these flags might not be necessary.
I started with an old configure line I had and worked on it until it
configured.)
I then ran 'make -j8' and got the usual slew of warnings that NAG
produces but then:
> [NAG Fortran Compiler normal termination, 4 warnings]
> MOD src/binding/fortran/use_mpi/mpi.mod-stamp
> NAG Fortran Compiler Release 6.1(Tozai) Build 6113
> [NAG Fortran Compiler normal termination]
> CCLD libch3affinity.la
> GEN lib/libmpi.la
> NAG Fortran Compiler Release 6.1(Tozai) Build 6113
> Option error: Unrecognised option -pthread
> make[2]: *** [lib/libmpi.la] Error 2
> make[2]: Leaving directory `/gpfsm/dswdev/mathomp4/MPI/src/mvapich2-2.2'
> make[1]: *** [all-recursive] Error 1
> make[1]: Leaving directory `/gpfsm/dswdev/mathomp4/MPI/src/mvapich2-2.2'
> make: *** [all] Error 2
In the immortal words of Dr Sam Beckett: "Oh boy".
This looks much like the problem that leads to me not being able to
build Open MPI with NAG:
> https://github.com/open-mpi/ompi/issues/1284
It's due to the fact that NAG uses -lpthread but gcc uses -pthread and
somehow the C threading flag gets pushed to NAG.
Does anyone know if there is something I can pass to MVAPICH2 to fix
this? Or perhaps do I need a patch?
Thanks,
Matt
--
Matt Thompson, SSAI, Sr Scientific Programmer/Analyst
NASA GSFC, Global Modeling and Assimilation Office
Code 610.1, 8800 Greenbelt Rd, Greenbelt, MD 20771
Phone: 301-614-6712 Fax: 301-614-6246
http://science.gsfc.nasa.gov/sed/bio/matthew.thompson
More information about the mvapich-discuss
mailing list