[Octopus-users] Troubles on opteron
ltu at fyslab.hut.fi
Mon Mar 7 08:12:15 WET 2005
I had a similar problem with the PGI compiler on opteron.
I guess your problems are in the out_inc.F90 file. Does octopus write
any wave functions? If not, the wave functions in subroutine
X(output_function) are not passed to the subroutine plain. Try changing
f(1:m%np) to f(:). This seems to work in the CVS version. The
segmentation fault probably comes in subroutine X(input_function). In
that case the reason would be the same and the similar change in the
Marin-Alin Elena wrote:
> I succeeded to compile and install octopus 1.4 on an opteron system.
> The troubles appear when I am running the Na2 file. I get a segmentation
> the configure command is
> ./configure --prefix=/r/alin/OCTOPUS
> --with-blas=/r/alin/ACML/pgi64/lib/libacml.a --with-gsl-prefix=/r/alin/GSL
> --with-fft=fftw3 --with-fft-lib="-L/r/alin/FFTW/lib/libfftw3f.a -lfftw3"
> LDFLAGS=-Bstatic FC=pgf90 CC=pgcc CXX=pgCC F77=pgf77 FCFLAGS="-O4 -fast
> FFTW is 3.0.3
> acml is 2.5
> gsl 1.6
> pg 5.2
> compiling it with the standard blas, lapack,gsl, fftw provided by suse 9.1
> distribution the problem is the same in the same point.
> I put in attach the inp and the log file in debug mode.
> the strangest thing is that on a 32 bits computer I had no problem with the
Lasse Tunturivuori The Gordian Maxim:
ltu at fyslab.hut.fi If a string has one end, it has another.
More information about the Octopus-users