On Fri, 2014-07-04 at 17:54 +0300, Mutlu COLAKOGULLARI wrote: > QUESTION: Why this difference is happening while the exact > parallelization is setted up as "-nk 1 -nd 1 -nb 1 -nt 1"?
because there are two different routines, one for serial diagonalization, the other for parallel diagonalization. The former is called if you explicitly set -nd 1, i.e. number of processors for linear algebra parallelization = 1; the latter is called if you do not specify a value for -nd because the code assumes -nd 4 (4 is the largest square integer contained in the number of processors, 8 in this case). You may argue "while do they use different outputs"? The answer is "for no good reason". Likely some changes to make the output more readable have been made in one routine but not in the other Paolo -- Paolo Giannozzi, Dept. Chemistry&Physics&Environment, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
