Dear wienkusers
I am trying to calculate the transport properties of Fe2MnCrSi
compound. While executing BoltzTrap I encounter some error like error
in factorization. May I kindly know how to solve this problem.
With regards
DP Rai
India
___
Hi,
Recently, I recompiled WIEN2k_13 with -O3 option. (I used to use the
default option O2). The full compiler option is as follows
-FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -traceback -assume
buffered_io -O3
I calculated the band structure of CsTlCl3 in high symmetry structure
(struct
Sadly, no surprise. Intel sometimes designs ifort for speed more than
accuracy, and they have done some things recently that can reduce accuracy.
Put -O3 immediately after -FR, the order matters. It may be better, or not.
On Aug 6, 2014 12:03 PM, James Peter jsqdmgl...@gmail.com wrote:
Hi,
Maybe try also:
-fltconsistency
and / or
-fp-model strict (or -fp-model precise might be enough)
I don't have all switches at hand, but there are more about floating point
operations found in the manual of the compiler
The warning of INTELL is:
Note
The options set by the O option may change
Maybe the BoltzTraP interpolation scheme (src/fite4.F90) has failed
because your k-mesh is too course. You could try increasing the number
of k-points to create a fine k-mesh (x kgen) followed by regenerating
case.energy[up/dn] (x lapw1 [-up/-dn])
5 matches
Mail list logo