[R-sig-hpc] Rmpi: -lopa parameter on compiling needed? Works without

[Arne Hendrik Schulz]

Dear Hao,
after getting some help from Cray directly, Rmpi compiles and works perfectly.
I had to comment out line 3786 in configure, so that there won't be any given parameter.
#PKG_LIBS="-L${MPI_LIBPATH} -lmpich_cray_default64 -lmpl -lopa ${MPI_LIBS}"
Additionally, I had to switch to the gnu-environement and build the C-Code using cc (Cray's own wrapper, cf [1]) instead of gcc. My whole command looks like this:
MAKEFLAGS="CC=cc" R2/bin/R CMD INSTALL Rmpi_0.6-4_cray.tar.gz --configure-args="--with-mpi=$MPICH_DIR --with-Rmpi-type=CRAY" --no-test-load

If you still run into error, you have to build R again using cc, CC and ftn instead of gcc, cpp and gfortran. Nevertheless, my setup worked like this.

Kind regards
Arne


[1] http://docs.cray.com/books/004-2179-001/html-004-2179-001/lymwlrwh.html
________________________________________
Von: Hao Yu <hyu at stats.uwo.ca>
Gesendet: Donnerstag, 17. April 2014 18:44
An: Arne Hendrik Schulz
Cc: r-sig-hpc at r-project.org
Betreff: Re: [R-sig-hpc] Rmpi: -lopa parameter on compiling needed? Works without

Hi Arne,

Last year, I was given an access of a Cray cluster to make Rmpi work.
After many tries and errors, -lmpich_cray_default64 -lmpl -lopa were used
eventually. Whether -lopa is necessary or not for other Cray cluster, I
have no answer. But it is good to know that some Cray clusters, -lopa is
not necessary.

Hao

Arne Hendrik Schulz wrote:
> Hi all,
> I am using the current development version of Rmpi (0.6-4) on a Cray
> Cluster. The problem is that it does not compile with the -lopa parameter
> as
> opa.c/libopa.c is not installed on the cluster (see attachted output
> below).
> Compiling Rmpi without the -lopa parameter works and using the cluster
> with
> MPI works, too. Is it really necessary to include libopa? I cannot see any
> drawbacks so far.
>
> Kind regards,
> Arne
>
> P.S.: If anyone would like to test without -lopa, I uploaded a branch of
> Rmpi-0.6-4: http://www.ahschulz.de/pub/R/packages/Rmpi_0.6-4_branch.tar.gz
> --with-Rmpi-type=CRAY uses the -lopa parameter,
> --with-Rmpi-type=CRAY_NOLOPA
> dies not.
>
> Installing Rmpi with -lopa:
> R/bin/R CMD INSTALL temp/packages/Rmpi_0.6-4.tar.gz
> --configure-args="--with-Rmpi-include=/opt/cray/mpt/6.1.1/gni/mpich2-cray64/
> 81/include
> --with-Rmpi-libpath=/opt/cray/mpt/6.1.1/gni/mpich2-cray64/81/lib
> --with-Rmpi-type=CRAY" --no-test-load
> * installing to library '/gfs1/work/hbmahs32/R/lib64/R/library'
> * installing *source* package 'Rmpi' ...
> checking for openpty in -lutil... no
> checking for main in -lpthread... no
> configure: creating ./config.status
> config.status: creating src/Makevars
> ** libs
> (...)
> gcc -std=gnu99 -shared -L/usr/local/lib64 -o Rmpi.so RegQuery.o Rmpi.o
> conversion.o internal.o -L/opt/cray/mpt/6.1.1/gni/mpich2-cray64/81/lib
> -lmpich_cray_default64 -lmpl -lopa
> /usr/bin/ld: cannot find -lopa
> collect2: error: ld returned 1 exit status
> make: *** [Rmpi.so] Error 1
> ERROR: compilation failed for package 'Rmpi'
>
>
> Installing Rmpi without -lopa
> R/bin/R CMD INSTALL temp/packages/Rmpi_0.6-4_branch.tar.gz
> --configure-args="--with-Rmpi-include=/opt/cray/mpt/6.1.1/gni/mpich2-cray64/
> 81/include
> --with-Rmpi-libpath=/opt/cray/mpt/6.1.1/gni/mpich2-cray64/81/lib
> --with-Rmpi-type=CRAY_NOLOPA" --no-test-load
> * installing to library '/gfs1/work/hbmahs32/R/lib64/R/library'
> * installing *source* package 'Rmpi' ...
> checking for openpty in -lutil... no
> checking for main in -lpthread... no
> configure: creating ./config.status
> config.status: creating src/Makevars
> ** libs
> (...)
> gcc -std=gnu99 -shared -L/usr/local/lib64 -o Rmpi.so RegQuery.o Rmpi.o
> conversion.o internal.o -L/opt/cray/mpt/6.1.1/gni/mpich2-cray64/81/lib
> -lmpich_cray_default64 -lmpl
> installing to /gfs1/work/hbmahs32/R/lib64/R/library/Rmpi/libs
> ** R
> ** demo
> ** inst
> ** preparing package for lazy loading
> ** help
> *** installing help indices
> ** building package indices
> * DONE (Rmpi)
>
> --
> Arne Hendrik Schulz
>
> ifib: Institute for Information Managment Bremen
> Am Fallturm 1 (entrance F)
> 28359 Bremen
> Germany
>
> Tel.: +49 (0)421 218 56585
> Fax: +49 (0)421 218 56599
> E-Mail: ahschulz at ifib.de
>
> Web: www.ifib.de
>
>
> _______________________________________________
> R-sig-hpc mailing list
> R-sig-hpc at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-hpc
>


--
Department of Statistics & Actuarial Sciences
Office Phone#:(519)-661-3622
Fax Phone#:(519)-661-3813
The University of Western Ontario
London, Ontario N6A 5B7
http://www.stats.uwo.ca/yu