Google found some info here:
https://software.intel.com/en-us/forums/topic/485528
You are mixing MKL libraries for the x86-64 target in R_LIB with MKL libraries
targeting the Intel Phi (MIC) co-processor in RP_LIB (those in
$MKLROOT/lib/mic/).
Are you really targeting the Intel Phi
You must run add -up and -dn to your command.
Regards,
--
Eamon
On Mar 28, 2014, at 10:15 AM, ben amara imen imen.benama...@gmail.com
wrote:
Hello,
I'm working on a magnetic compound . I do the SCF (runsp_lapw) and it went
well. Then to calculate the DOS, I executed the command
Crystallography.nethttp://Crystallography.net appears to have the structure
you want: http://www.crystallography.net/9011965.html
Simply download the .CIF file and use 'cif2struct'.
Regards,
--
Eamon McDermott, M.Sc.
Institute of Materials Chemistry, TU Wien
Then the problem is exactly as Tomas has described it. If lapw1 is searching
for a .vsp file (and not .vspup/.vspdn) then you are running 'run_lapw' when
you have initialized for a spin-polarized case.
I agree that you should re-start your calculation from a clean directory with
only the
no space left on device
You have run out of disk space on the device containing your output directory,
or for some other reason the process is unable to write to the disk (such as
NFS issues).
Best regards,
--
Eamon McDermott, M.Sc.
Institute of Materials Chemistry, TU Wien
I recommend checking for NaN's in your Cu.scf0 file.
This behaviour looks similar to what I've seen before in some cases when
using the broken FFT library in the early 12.1 release. It can be fixed
for the serial mode lapw0 by adding -DFFTW3 or -DFFTW2 to FOPT and
linking to an fftw in R_LIBS.
6 matches
Mail list logo