Unexpected crush at the dipoles calculation with yambo5.2

Various technical topics such as parallelism and efficiency, netCDF problems, the Yambo code structure itself, are posted here.

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

Post Reply
User avatar
jasonhan0710
Posts: 60
Joined: Wed Dec 23, 2020 6:48 am
Location: China

Unexpected crush at the dipoles calculation with yambo5.2

Post by jasonhan0710 » Wed Sep 27, 2023 2:51 am

Dear all,

I have updated the yambo to version 5.2. However, after I compiled it with Intel oneapi, it stopped every time and without any error information when calculating the dipole term in GW. Would you please help me figure out this problem? Thank you!

Best,
Jason
Jason Han

Assistant Professor
Department of Physics
National University of Defense Technology
Hunan, China

User avatar
Nicola Spallanzani
Posts: 65
Joined: Thu Nov 21, 2019 10:15 am

Re: Unexpected crush at the dipoles calculation with yambo5.2

Post by Nicola Spallanzani » Wed Sep 27, 2023 8:07 am

Dear Jason,
could you provide more information? For example sending the output and error files, the files "config.log", "config/report" and one of the log file that you can find in the LOG/ directory.

Best,
Nicola
Nicola Spallanzani, PhD
S3 Centre, Istituto Nanoscienze CNR and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu

User avatar
jasonhan0710
Posts: 60
Joined: Wed Dec 23, 2020 6:48 am
Location: China

Re: Unexpected crush at the dipoles calculation with yambo5.2

Post by jasonhan0710 » Wed Sep 27, 2023 9:21 am

Hi Nicola,

Thanks for your reply.

The config.log file is attached.

I set

Code: Select all

./configure --enable-open-mp --enable-mpi \
	    --enable-time-profile --enable-memory-profile --enable-msgs-comps \
	    --enable-slepc-linalg \
	    --with-blas-libs="${MKL_LIBS}" --with-lapack-libs="${MKL_LIBS}" \
	    --with-fft-includedir="${MKLROOT}/include" --with-fft-libs="${MKL_LIBS}" \
        --enable-dp
to compile the code.

The GW calculation stops at

Code: Select all

 <02s> P1-c55: [DIP] Database not correct or missing. To be computed
 <05s> P1-c55: [x,Vnl] computed using 130 projectors
 <05s> P1-c55: [WARNING] [x,Vnl] slows the Dipoles computation. To neglect it rename the ns.kb_pp file
 <05s> P1-c55: Dipoles: P, V and iR (T) |                                        | [000%] --(E) --(X)
and the report files shows that

Code: Select all

[06] Dipoles
 ============

 [RD./SAVE//ns.kb_pp_pwscf]------------------------------------------------------
  Fragmentation                                    : yes
 - S/N 002730 ---------------------------------------------- v.05.01.00 r.22170 -

 [WARNING] [x,Vnl] slows the Dipoles computation. To neglect it rename the ns.kb_pp file
without and error information.

The slurm reports

Code: Select all

===================================================================================
=   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
=   RANK 2 PID 17140 RUNNING AT c56
=   KILLED BY SIGNAL: 11 (Segmentation fault)
===================================================================================
The same case can be calculated successfully with yambo-5.0.4.

Thank you!

Best,
Jason
You do not have the required permissions to view the files attached to this post.
Jason Han

Assistant Professor
Department of Physics
National University of Defense Technology
Hunan, China

User avatar
claudio
Posts: 463
Joined: Tue Mar 31, 2009 11:33 pm
Location: Marseille
Contact:

Re: Unexpected crush at the dipoles calculation with yambo5.2

Post by claudio » Wed Sep 27, 2023 10:32 am

Dear Jason

you can try to compile with debug flags, just add --enable-debug-flags to the configure,
and see if you get more information about the crash

another advice: you may remove --enable-dp
usually double precision is not so important for the GW calculations


best
Claudio
Claudio Attaccalite
[CNRS/ Aix-Marseille Université/ CINaM laborarory / TSN department
Campus de Luminy – Case 913
13288 MARSEILLE Cedex 09
web site: http://www.attaccalite.com

User avatar
jasonhan0710
Posts: 60
Joined: Wed Dec 23, 2020 6:48 am
Location: China

Re: Unexpected crush at the dipoles calculation with yambo5.2

Post by jasonhan0710 » Thu Sep 28, 2023 9:09 am

Dear Claudio and Nicola,

Thank you for your quick reply. I have recompiled yambo code with "--enable-debug-flags", however, it dose not give any error information in the report file and log file. The report file stops as previously and the log file shows that

Code: Select all

<---> [01] MPI/OPENMP structure, Files & I/O Directories
 <---> MPI Cores-Threads   : 1(CPU)-40(threads)
 <---> [02] CORE Variables Setup
 <---> [02.01] Unit cells
 <---> [02.02] Symmetries
 <---> [02.03] Reciprocal space
 <---> [02.04] K-grid lattice
 <---> Grid dimensions      :  20  20
 <---> [02.05] Energies & Occupations
 <---> [03] Transferred momenta grid and indexing
 <---> [04] Dipoles
 <02s> [DIP] Database not correct or missing. To be computed
 <03s> [x,Vnl] computed using 130 projectors
 <03s> [WARNING] [x,Vnl] slows the Dipoles computation. To neglect it rename the ns.kb_pp file
 <03s> Dipoles: P, V and iR (T) |                                        | [000%] --(E) --(X)[1]    48147 segmentation fault 
Do you have any thoughts on the problem?

Best,
Jason
Jason Han

Assistant Professor
Department of Physics
National University of Defense Technology
Hunan, China

User avatar
claudio
Posts: 463
Joined: Tue Mar 31, 2009 11:33 pm
Location: Marseille
Contact:

Re: Unexpected crush at the dipoles calculation with yambo5.2

Post by claudio » Thu Sep 28, 2023 10:50 am

Dear Jason

it is difficult to day, where is the problem.
Another test you can do it is to remove the MKL libraries

./configure --enable-open-mp --enable-mpi --enable-slepc-linalg

and try to run it again
Claudio
Claudio Attaccalite
[CNRS/ Aix-Marseille Université/ CINaM laborarory / TSN department
Campus de Luminy – Case 913
13288 MARSEILLE Cedex 09
web site: http://www.attaccalite.com

User avatar
Daniele Varsano
Posts: 3926
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Unexpected crush at the dipoles calculation with yambo5.2

Post by Daniele Varsano » Sat Sep 30, 2023 6:22 am

Dear Jason,
besides what suggested by Claudio, it is useful to know the setup of your calculation. It is possible you are running out of memory.
You can add in the configure procedure the flag --enable-memory-profile and post your input and log 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/

User avatar
jasonhan0710
Posts: 60
Joined: Wed Dec 23, 2020 6:48 am
Location: China

Re: Unexpected crush at the dipoles calculation with yambo5.2

Post by jasonhan0710 » Sun Oct 01, 2023 4:03 am

Hi Daniele and Claudio,

Thanks for helping me solve the problem. After I removed the MKL libraries, yambo can run normally. Thank you!

Best,
Jason
Jason Han

Assistant Professor
Department of Physics
National University of Defense Technology
Hunan, China

Post Reply