# !!! Please do not send me any queries on how to install MPI, I will # options: -Msecond_underscore (compilation) and -g77libs (linking) # compiled with f77/g77, but then you need to add the following # please note that you might be able to use a lam or mpich version # band with of roughly 160 MBit/s (full duplex) # lam was generally faster and we found an average communication configure -prefix /usr/local/lam-6.5.X -with-cflags=-O -with-fc=pgf90 \ # We found that mpich.1.2.1 and lam-6.5.X are stable # Precompiled mpi version will also not work !!! # compiler however appends only one underscore. # appends *two* underscores to symbols that contain already an # You must *not* compile mpi with g77/f77, because f77/g77 # MPI section, uncomment the following lines # please do not send us any querries reltated to FFTW (no support) # if you use fftw3d you need to insert -lfftw in the LIB line as well # since the FFTW is very slow for radices 2^n the fft3dlib is used vasp.5.lib/linpack_double.o -L/usr/lib64 -llapack vasp.5.lib/linpack_double.o $(LAPACK) \ vasp.5.lib/lapack_atlas.o -llapack -lcblas # LAPACK, simplest use vasp.5.lib/lapack_double # use the mkl Intel libraries for p4 ( #BLAS=-L/opt/intel/mkl/lib/32 -lmkl_p4 -lpthread # use specific libraries (default library path points to other libraries) #ATLASHOME= $(HOME)/archives/BLAS_OPT/ATLAS/lib/Linux_ATHLONXP_SSE1/ # Athlon: Atlas based BLAS are presently the fastest # P4: VASP works fastest with Intels mkl performance library # what you chose is very system dependent # the following lines specify the position of BLAS and LAPACK # (even on Athlon system, Athlon specific optimistation worsens performance) # the performance, and found no improvements with -O3-5 or -fast # we have tested whether higher optimisation improves # the option will not do any harm on other 3.X pgf90 distributionsįFLAGS = -ffree-form -ffree-line-length-none # the -Mx,119,0x200000 is required if you use older pgf90 versions # general fortran flags (there must a trailing blank on this line) Dkind8 -DNGXhalf -DCACHE_SIZE=2000 -DGfortran -Davoidalloc \ # **** definitely use -DRACCMU_DGEMV if you use the mkl library # RACCMU_DGEMV use DGEMV instead of DGEMM in RACC (faster on P4) # RPROMU_DGEMV use DGEMV instead of DGEMM in RPRO (usually faster) # wNGXhalf gamma point only reduced in X direction # NGXhalf charge density reduced in X direction # SUSE 6.X, maybe some Red Hat distributions: # that's probably the right line for some Red Hat distribution: # that's the location of gcc for SUSE 5.3 # whereis CPP ? (I need CPP, can't use gcc with proper options) # all CPP processed fortran files have the extension. # routines are not optimal on the Athlon. # you certainly need atlas on the Athlon, since the mkl # 3b) or obtain the atlas based BLAS routines # this is really excellent when you use Intel CPU's # 3a) Intels own optimised BLAS (PIII, P4, Itanium) # the two most reliable packages around are presently: # 2) most desirable: get an optimized BLAS # ZHEEV (however this was with lapack 1.1 now I use lapack 2.0) # use pgf77 or pgf90 for BLAS, VASP hang up when calling # please use g77 or f77 for the compilation. # and compile the blas routines (BLAS/SRC directory) # is therefore wrong by about 1meV/atom). # libm causing small errors in the error-function (total energy # Mind that some Linux distributions (Suse 6.1) have a bug in # Hence check ***ALL**** options in this makefile very carefully # it might be required to change some of library pathes, since # The makefile was tested only under Linux on Intel platforms # any of the releases, even Athlon or PIII optimisation does # we have found no noticable performance differences between # Makefile for Portland Group F90/HPF compiler release 3.0-1, 3.1 # rewritten for gfortran under Redhat, RedHat on x86_64 and gfortran The Makefiles are for a build of vasp that includes the climbing-image NEB add-ons from University of Austin at Texas theory group (see ).įrank did a little benchmarking using the Hg job provided with VASP4: the original makefile for linux had lots of things missing and some problems with trailing EOL's and spaces. These Makefiles were created by our systems manager, Frank Bures. In case these are of use to you (or anyone else), here are two Makefiles for vasp on an x86_64 linux machine, running RedHat.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |