Machine specific configure scripts

From The Yambo Project
Jump to: navigation, search

Preinstalled

The GPL version of Yambo is already installed on different HPC systems around the world, here we report some of them: Niflheim at Technical University of Denmark SP6 at CINECA Arina at SGI at Universidad del Pais Vasco. Core.Sam at University of Pittsburgh.

HPCs

Below are some configure options that have been used in the past. Of course, since compilers and architectures vary a lot, there are no guarantees that they will work on your system. Be particularly careful when specifying FCFLAGS, as you may override settings which are necessary for compilation, e.g. -nofor_main with ifort.

CH: Suggest to list/group these in order of architecture, linux, IBM, Cray, OS/X, etc

HLRN

EURORA @ CINECA

EURORA is a hybrid supercomputer, with Intel Xeon andyBridge processors and GPU NVIDIA Tesla K20 accelerators

module load autoload/0.1 intel/cs-xe-2013--binary intelmpi/4.1.0--binary mkl/11.0.1--binary gnu/4.6.3 cuda/5.0.35 qe/5.0.3 netcdf/4.1.3--intel--cs-xe-2013--binary hdf5/1.8.9_ser--intel-
-cs-xe-2013--binary szip/2.1--gnu--4.6.3 zlib/1.2.7--gnu--4.6.3
./configure --with-p2y=5.0 \
--with-iotk=/cineca/prod/build/applications/qe/5.0.3/cuda--5.0.35/BA_WORK/espresso-5.0.3/iotk/ \
--with-netcdf-lib=/cineca/prod/libraries/netcdf/4.1.3/intel--cs-xe-2013--binary/lib/ \
--with-netcdf-include=/cineca/prod/libraries/netcdf/4.1.3/intel--cs-xe-2013--binary/include \
--with-netcdf-link="-L/cineca/prod/libraries/hdf5/1.8.9_ser/intel--cs-xe-2013--binary/lib -L/cineca/prod/libraries/szip/2.1/gnu--4.6.3/lib -lhdf5_fortran -lhdf5_hl -lhdf5 -lnetcdff -lnetcdf -lcurl -lsz -lz"

IBM AIX and xlf (SP6, CINECA)

Linking with netCDF, PWscf, FFTW. Production runs

export CPP=cpp 
export CC=xlc_r
export F77=xlf_r
export FC=xlf90_r
export FCFLAGS='-O2 -q64 -qstrict -qarch=pwr6 -qtune=pwr6 -qmaxmem=-1 -qsuffix=f=f'
./configure --build=powerpc-ibm --with-fftw=/cineca/prod/libraries/fftw/3.2.2/xl--10.1/lib 
--with-netcdf-lib=/cineca/prod/libraries/netcdf/4.0.1/xl--10.1/lib 
--with-netcdf-include=/cineca/prod/libraries/netcdf/4.0.1/xl--10.1/include 
--with-iotk=/cineca/prod/build/applications/QuantumESPRESSO/4.1/xl--10.1/BA_WORK/QuantumESPRESSO-4.1/iotk 
--with-p2y=4.0

To compile, use GNU make: ./gmake yambo interfaces GNU/Linux: ifort

ifort 11.0 with netCDF and iotk/p2y 4.0 support, serial ./configure FC=ifort --with-netcdf-lib=/usr/local/libraries/netcdf/4.0.1/ifort--11.1/lib --with-netcdf-include= /usr/local/libraries/netcdf/4.0.1/ifort--11.1/include --with-iotk=/opt/espresso/4.1.2/iotk/ --with-p2y=4.0 --with-blacs=no BLAS, LAPACK and FFTW linked using the mkl libraries

The correct linking path can be found here (thanks to Giovanni Pizzi for this!!!). A typical configure command line that can be used to link the mkl BLAS, LAPACK and FFTW is

./configure FC=ifort \
--with-blas="-L/opt/intel/Compiler/11.1/073/mkl/lib/em64t/ -lmkl_core -lmkl_intel_lp64 -lmkl_sequential" \
--with-lapack="-L/opt/intel/Compiler/11.1/073/mkl/lib/em64t/ -lmkl_core -lmkl_intel_lp64 -lmkl_sequential" \
--with-blacs="-L/opt/intel/Compiler/11.1/073/mkl/lib/em64t/ -lmkl_core -lmkl_intel_lp64 -lmkl_sequential" \
--with-scalapack="-L/opt/intel/Compiler/11.1/073/mkl/lib/em64t/ -lmkl_core -lmkl_intel_lp64 -lmkl_sequential" \
--with-fftw="/opt/intel/Compiler/11.1/073/mkl/lib/em64t/" \
--with-fftw-lib="-lmkl_core -lmkl_intel_lp64 -lmkl_sequential" \
 

Production runs

FCFLAGS='-O3 -xW -tpp7 -assume bscc -nofor_main' LDFLAGS='-xW' --with-blacs=no ./configure FC=mpif90 Debugging runs

./configure --enable-debug FC=mpif90 CFLAGS='-O0 -g' FCFLAGS='-O0 -tpp7 -C -check noarg_temp_created -w90 -w95 -assume bscc -nofor_main -g' --with-blacs=no GNU/Linux: g95

Debugging runs

./configure FC=g95 FCFLAGS='-O0 -fbackslash -g' CFLAGS='-O0 -g -Dextcus -Dextfus' CC=gcc --with-blacs=no Production runs

./configure FC=g95 FCFLAGS="-O3 -fbackslash -fno-second-underscore" GNU/Linux: gfortran

Debugging runs

./configure --enable-debug --without-mpi FC=gfortran FCFLAGS='-O0 -g -fbounds-check' CFLAGS='-O0 -g -Dextcus -Dextfus' CC=gcc --with-blacs=no GNU/Linux: PGI Fortran

./configure FC=pgf95 GNU/Linux: SUN Fortran

./configure FC=f95 GNU/Linux: Pathscale Fortran

./configure FC=pathf90 PFC=mpif90 FCFLAGS="-fno-second-underscore" UFFLAGS="-fno-second-underscore" GNU/Linux: Open64 Fortran

./configure FC=openf95 FCFLAGS="-fno-second-underscore" UFFLAGS="-fno-second-underscore" GNU/Linux using xlf

Tested on Mare nostrum (ppc64) Production runs

./configure CC=xlc F77=xlf PFC=mpif90 --with-blacs=no with NETCDF support

./configure CC=xlc F77=xlf PFC=mpif90 \

--with-netcdf-include=/gpfs/apps/NETCDF/netcdf-3.6.0_64/include \
--with-netcdf-lib=/gpfs/apps/NETCDF/netcdf-3.6.0_64/lib --with-blacs=no

Magerit (CeSViMa)

Production runs with NetCDF support

./configure PF90=xlf90_r FC=xlf90_r F77=xlf_r CC=xlc_r CPP="cpp -P" \ --with-blas=/gpfs/apps/BLAS/1.0.0/64/lib/libblas.a --with-lapack=/gpfs/apps/LAPACK/3.0/64/liblapack.a \ --with-netcdf-include=/gpfs/apps/NETCDF/3.6.0/64/include --with-netcdf-lib=/gpfs/apps/NETCDF/3.6.0/64/lib \ --with-fftw=/gpfs/apps/FFTW/3.2/64/lib Default

./configure ACML libraries

Production runs

./configure --with-blas=$ACMLPATH/libacml.a --with-lapack=$ACMLPATH/libacml.a FFT in ACML not supported yet MKL libraries

Production runs

export MKLROOT=/applis/intel/mkl export MKLPATH=${MKLROOT}/lib/intel64 export MKLINCLUDE=${MKLROOT}/include export MKLLIBS="-L${MKLPATH} -I${MKLINCLUDE} -I${MKLINCLUDE}/intel64/lp64 -llapack -lmkl_blas95_lp64 -lmkl_lapack95_lp64 -Wl,--start-group

               ${MKLPATH}/libmkl_intel_lp64.a ${MKLPATH}/libmkl_sequential.a ${MKLPATH}/libmkl_core.a -Wl,--end-group -lpthread"

./configure FC=ifort F77=ifort --with-blas="$MKLLIBS" --with-lapack="$MKLLIBS" use this page to determine the exact library version for your system: Intel MKL Link Line Advisor FFT in MKL are supported through the FFTW interface presents in the latest MKL packages. --with-fftw="${MKLROOT}/interfaces/fftw3xf/" --with-fftw-lib="${MKLROOT}/interfaces/fftw3xf/libfftw3xf_intel.a $MKLLIBS" Copyright ©2008-2017 Andrea Marini