BSE in serial crash for large matrix

You can find here problems arising when using old releases of Yambo (< 5.0). Issues as parallelization strategy, performance issues and other technical aspects.

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

Locked
javad
Posts: 44
Joined: Fri Feb 28, 2014 10:23 pm

BSE in serial crash for large matrix

Post by javad » Wed Mar 23, 2016 11:23 am

Dear Yambo developpers

I tried to run yambo for a system with many k-points, and few bands.

Since it takes only few hours, I sent the job in serial. The problem
is that if the matrix is too large I get:
"...

<04h-25m-10s> Kernel |#################################### | [090%] 04h-13m-23s(E) 04h-40m-26s(X)
<04h-28m-54s> Kernel |##################################### | [092%] 04h-17m-08s(E) 04h-37m-35s(X)
<04h-32m-36s> Kernel |###################################### | [095%] 04h-20m-49s(E) 04h-33m-52s(X)
[ERROR] STOP signal received while in :[06.04.03] Kernel loop
[ERROR][NetCDF] NetCDF: One or more variable sizes violate format constraints
... ¨

regards
Javad
Javad Exirifard
IPM - Institute for Research in Fundamental Sciences
P. O. Box 19395-5746
Niavaran Square
Tehran, Iran

User avatar
Daniele Varsano
Posts: 4209
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: BSE in serial crash for large matrix

Post by Daniele Varsano » Wed Mar 23, 2016 2:18 pm

Dear Javad,
this is due to the netcdf format which has a limited number of variables.
The problem can be solve by avoiding to write the BSE matrix on disk by setting in input:

Code: Select all

DBsIOoff= "BS" 
In general by adding -V io in your command line you will have variables governing IO in the input file.

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/

Locked