Dear Carlo,
I do not have direct experience nor with QE, but I have seen this from
another code:
--------------------------------------------------------------------------
Date: Fri, 27 Jan 2017 14:37:06 +0000
From: Iain Bethune <[email protected]>
Reply-To: [email protected]
To: [email protected]
Subject: Re: [CP2K:8598] error in qs_ks_methods.F depending on number of cores
(intel compiler)
Further to what I reported this morning, I found that the MKL release
2017.1.132 is buggy. I was able to get a successful build using the compiler
from the 2017.1.132 release and MKL from the previous (2017.0.098) release.
--------------------------------------------------------------------------
Personally I had similar issues (again with CP2K), but I do not remember
if it was with exactly that version; in any case I reverted from Intel
v2017 to v2017 and I was able to run the code.
Maybe not very useful, but... I would try with 'gfortran' (and OpenMPI,
if possible).
Greetings,
apsi
-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
Ari Paavo Seitsonen / [email protected] / http://www.iki.fi/~apsi/
Ecole Normale Supérieure (ENS), Département de Chimie, Paris
Mobile (F) : +33 789 37 24 25 (CH) : +41 79 71 90 935
On Fri, 17 Mar 2017, Carlo Nervi wrote:
Dear QE community,a colleague of mine is using Intel compiler and libraries to
compile QE 6.1.
We used
compilers_and_libraries_2017.1.132
and
parallel_studio_xe_2017.1.043
I got apparently normal results using "mpirun -np 21", but very different results using
"mpirun -np 21 -nk 7", and relevant longer time of
calculations.
I suspect there is something wrong with libraries or mpi.
Before to dig more deep in the debugging, do anybody is aware of any
inconsistencies/anomalies using the mentioned compiler/libraries?
We are using also docker...
Thank you,
Carlo
--
------------------------------------------------------------
Prof. Carlo Nervi [email protected] Tel:+39 0116707507/8
Fax: +39 0116707855 - Dipartimento di Chimica, via
P. Giuria 7, 10125 Torino, Italy. http://lem.ch.unito.it/
_______________________________________________
Pw_forum mailing list
[email protected]
http://pwscf.org/mailman/listinfo/pw_forum
