Dear prof. Blaha,
I tried the new version of lapwso. I replaced mine and recompiled it
with no problems, however this did not solve the problem. I'm still
getting the same error. It definitely has to do with mpi, because serial
version works fine (but takes ages for a single iteration).
Duri
Thank you for your answer and suggestions.
I tried the same calculation with a ridiculous amount of memory (200GB, while
non so calculations finish easily on 10 GB), and a recompiled lapwso. Nothing
changed unfortunately...
I'm lost now, and have no idea how to approach the problem, so
29.06.2016 22:07, Maciej Polak wrote:
I'm curious though, how does it get there?
I am not a specialist in commands of shell, but there are several
commands writing their output to .timeso in lapwsopara, for example:
(cd $PWD;$t $exe ${def}_${loop}.def;rm -f .lock_$lockfile[$p])
>>.timeso_$loop
Dear Lyudmila,
Thank you for the reply. I agree that this looks like LAPACK related
stuff. I'm curious though, how does it get there?
My case.inso and case.struct are fine I hope (case.struct is an effect
of geometry optimization, and it looks fine in xcrysden and the forces
are small). See
29.06.2016 12:11, Maciej Polak wrote:
I have trouble with running run_lapw with the "so" switch in parallel.
I narrowed down the origin of this error to be located in .timeso_X
files (.timeso_1).
info in pzheevx16 5 2 1
{1,0}: On entry to
PZUNMTR parameter number
Dear WIEN2k Developers and Users,
I have trouble with running run_lapw with the "so" switch in parallel.
What I get is an error:
Missing }.
> stop error
I narrowed down the origin of this error to be located in .timeso_X
files (.timeso_1).
From a brief look at the lapwsopara_lapw script
6 matches
Mail list logo