[CPMD-list] Compiling cpmd2cube on BG/L

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Mon Apr 23 20:53:33 CEST 2007


On Mon, 23 Apr 2007, Matteo Guglielmi wrote:

hi matteo,

MG> I'm having troubles compiling cpmd2cube on BG/L (EPFL, Lausanne)
MG> 
MG> Actually I get it compiled using this header:
MG> 
MG> #--------------- Configuration for IBM-BGL-FEN --------------
MG> FC      = xlf90 -O3 -qstrict  -qarch=auto -qxlf77=leadzero
MG> FFLAGS  = -WF,"-D__PWRLinux,-DFFT_FFTW" -qsuffix=cpp=F -qmaxmem=32768
MG> LFLAGS  = -O3
MG> LIBS    = -L/bgl/fftw-2.1.5/lib -lsfftw

this is the problem!
you are using a library for the cpus on the BG/L nodes,
those are quite different from the ones on the frontend
machine. you have to roll your own fftw2 (and the double
precision version at that). that is why i have:
LIBS    = -L$(HOME)/fftw-2.1.5/lib -ldfftw

in my config.

ciao,
  axel.

MG> # uncomment and set only if needed
MG> #CC     = cc
MG> #CFLAGS =
MG> #--------------- End of Configuration ---------------------------------
MG> 
MG> but when I run it I get this error message:
MG> 
MG> #---------------------
MG> [bg-fe01:/bglscratch/guglielm/GLYTRPH/RUN1/KS] > cpmd2cube.x WAVEFUNCTION.49
MG>  
MG> Illegal instruction
MG> #----------------------
MG> 
MG> 
MG> Thanks,
MG> MG.
MG> 
MG> 
MG> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.



More information about the CPMD-list mailing list