Response calculation is stuck

Deals with issues related to computation of optical spectra in reciprocal space: RPA, TDDFT, local field effects.

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano, Conor Hogan

Post Reply
DavidPolito93
Posts: 71
Joined: Sat Jun 06, 2020 10:43 am

Response calculation is stuck

Post by DavidPolito93 » Tue Nov 29, 2022 4:06 pm

Dear all,

I am studying a 2D system. This system has spin-orbit interactions. I am using the RIM corrections and the box cutoff for the Coulomb potential in the non-periodic direction. I am interested in the GW corrections. I am having some issues in the calculation of the dynamical matrix in PPA. I am running in parallel. I am using Yambo Version 5.1.0.

More precisely, after finishing the calculation of Xo@q[1] it gets stuck at X@q[1] : it seems that it has finished but it is not progressing after.

I am attaching the input, the report and the LOG file for the first CPU.

Thank you in advance,

Davide Romanin
You do not have the required permissions to view the files attached to this post.
-----------------------------------------------------
Assistant Professor
Polytech - Paris-Saclay University
C2N, CNRS
10 Bd Thomas Gobert
91120 Palaiseau

DavidPolito93
Posts: 71
Joined: Sat Jun 06, 2020 10:43 am

Re: Response calculation is stuck

Post by DavidPolito93 » Wed Nov 30, 2022 1:01 pm

Dear all,

just an update : if I try with only 2 CPUs it works (even if it goes very slowly as expected), however as soon as I increase the number of processor it gets stuck after computing X0.

It seems that only two processors have a workload, while the others don't. Indeed in the LOG files from CPUS>2 (i.e. 3,4,...) I get a warning telling me that that CPU doesn't have things to do.

Do you have any suggestion on how I could fix this?

Best,

Davide Romanin
-----------------------------------------------------
Assistant Professor
Polytech - Paris-Saclay University
C2N, CNRS
10 Bd Thomas Gobert
91120 Palaiseau

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Response calculation is stuck

Post by Daniele Varsano » Wed Dec 07, 2022 11:15 am

Dear Davide,

tha fact ncpu > 2 has a zero workload, it is a normal behaviour when inverting the X, being a PP you have only 2 frequencies unless you use a parallel linear algebra (scalapack).
Having saying that, I do not understand why the code stops.
What you can try is to assign an explicit parallelization strategy in input, e.g.:

Code: Select all

X_and_IO_CPU= "1 1 1 12 2"                 # [PARALLEL] CPUs for each role
X_and_IO_ROLEs= "q g k c v"               # [PARALLEL] CPUs roles (q,g,k,c,v)
Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Post Reply