C++ Python C Makefile Perl Fortran Other
Latest commit af09913
Feb 17, 2017
Also added JSON output for parameter derivatives.
|Failed to load latest commit information.|
To build: for the impatient: cd src; make; for the slightly less impatient: cd src; make PLATFORM=ARCH; NOTE: The build system currently requires GNU make. This can usually be called as 'make' or 'gmake'. There are several makefile stubs in make/. You can look for your architecture there or roll your own. Typing 'make' (or possibly gmake) without any arguments will cause the build program to look for `uname -s`.mk(for example, on Linux, it will look for Linux.mk). You can choose the platform by typing 'make PLATFORM=arch'. There are various variables to set in the .mk files. The most important are CXX and CXXFLAGS; the rest are performance and debugging-related(these are also very important in large calculations!) See make/Linux.mk for an example of a fairly well-optimized .mk file. Makefile variables: CXX: C++ compiler. For example, for a serial compile, set this to g++. For parallel, set it to mpiCC or mpicxx (or mpic++). CXXFLAGS: Flags to pass to the compiler. Put optimization (-O2) flags, etc here. Must also include the option "$(INCLUDEPATH)" DEBUG: Define any debugging flags here. An optimized DEBUG might be '-DNDEBUG'. A safe and slow DEBUG could be '-DDEBUG_WRITE -DRANGE_CHECKING' LDFLAGS: Linker flags. Usually you don't need anything here, but some compilers need '-lm' for the math libraries. BLAS_LIBS: The BLAS libraries(for example: -L/usr/lib -lcblas LAPACK_LIBS: LAPACK libs(as BLAS) (ex. -L/opt/lapack/lib -llapack) LAPACK_INCLUDE: LAPACK headers (for examples -I/opt/lapack/ EINSPLINE_LIBS: Einspline libs EINSPLINE_INCLUDE: Einspline include directory DEPENDMAKER: If you have gcc, it should be g++ -MM -I $(INCLUDEPATH). List of preprocessor flags(these may be defined by -D[FLAG] with most compilers. List them in the CXXFLAGS variable along with any optimization flags.) DEBUG_WRITE: Enable extra debugging output. RANGE_CHECKING: Enable range checking of arrays. Big performance hit if you turn this on! NDEBUG: Turn off assert()'s in code. For less safety and better performance set this flag. USE_BLAS: Enable usage of CBLAS libraries. This will enable BLAS in some of the linear algebra routine and enable BLAS_MO, which is a very fast MO evaluator for some archetectures(Itanium is one that really benefits from this). You must also set the BLAS_LIBS and BLAS_INCLUDE variables. USE_LAPACK: Enables usage of LAPACK libraries in pw2lcao. It's not used in the main program, but improves pw2lcao's performance by several orders of magnitude. You must also set the LAPACK_LIBS and BLAS_INCLUDE variables. USE_MPI: Enable use of MPI parallelization. For large calculations, this is quite necessary. USE_EINSPLINE: Use the einspline libraries from Ken Esler to evaluate blips. EINSPLINE_LIBS and EINSPLINE_INCLUDE must also be set.