Compilers

Please use IBM compilers to create an optimized program from source code. The following compilers are available: - XL C/C++ (C/C++ compiler) version 13.1 - XLF Fortran version 15.1 GNU compilers are also available, but it is recommended to try XL compilers first.

Environment setup

• IBM C/C++/fortran - compilers are available with standard settings

• GNU C/C++/fortran - compilers are available with standard settings

64-bit addressing

Normally, compilers operate in 32 bit mode. You can compile your programs in 64 bit mode to achieve a higher performance as well as use more memory. You can do this by using the “-q64” flag. You can also set the variable OBJECT_MODE using this command:

export OBJECT_MODE=64

When using GNU compilers use “-maix64” flag.

Serial Code Compilation

Optimized code compilation examples:

C:xlc_r -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.c

C++:xlC_r -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.c

Fortran:xlf_r -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.f

Fortran90:xlf90_r -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.f90

You can find a more detailed description of IBM compilers using -qhelp flag or here: http://publib.boulder.ibm.com/infocenter/comphelp/v101v121/index.jsp

OpenMP Code Compilation

Optimized code compilation examples using Open Multi-Processing (OpenMP):

C:xlc_r -q64 -qsmp=omp -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.c

C++:xlC_r -q64 -qsmp=omp -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.c

Fortran:xlf_r -q64 -qsmp=omp -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.f

Fortran90:xlf90_r -qsmp=omp -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.f90

IBM Parallel Environment

Optimized code compilation examples using MPI on Aurel supercomputer:

C:mpcc -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.c

C++:mpCC -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.c

Fortran:mpxlf -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot _myprog.f

Fortran90:mpxlf90 -q64 -qarch=pwr7 -qtune=pwr7 -O3 -qhot myprog.f90

MPICH2

You can load the necessary command through modules (module load mpi/mpich2). The syntax is the same as for IBM PE, only the commands themselves are different:

C:mpicc

C++:mpic++

Fortran 77: mpif77

Fortran 90: mpif90