Estimating the run time for getting optical absorption

[niloufar]

Hello dear developers,

I am running the Yambo code for a 40-atom cell. As a first step I am trying to get the optical absorption and I use 12 cores in my computer, but after 2 days it hasn't made the output folder yet! and sticks in the stage of dipoles! Is it normal or something is wrong? (By the way its nscf took 4 days!)

Here are the last lines in my output now (from 2 days ago):

[04] Dipoles
============


[WARNING] DIPOLES database not correct or not present
[RD./SAVE//ns.kb_pp_pwscf]----------------------------------
Fragmentation :yes
- S/N 007450 -------------------------- v.04.05.03 r.09122 -

[WARNING] [x,Vnl] slows the Dipoles computation. To neglect it rename the ns.kb_pp file
[WF-Oscillators/G space loader] Normalization (few states) min/max :0.982E-20 1.00



Thank you in advance @>-

[Daniele Varsano]

Dear Niloufar,
it really depends on the size of your system and the number of bands you included.
Anyway you can speed up the calculation of dipoles by neglecting the non local commutator as indicated in the report file:

Code: Select all

[WARNING] [x,Vnl] slows the Dipoles computation. To neglect it rename the ns.kb_pp file

so you need to rename the ns.kb_pp for instance:
mv ./SAVE/ns.kb_pp_pwscf ./SAVE/ns.kb_pp_pwscf_

Next you can also optimise the parallelism strategy in the input, as fo instance:

Code: Select all

DIP_CPU= " 1 12 2"                      # [PARALLEL] CPUs for each role
DIP_ROLEs= "k c v"   

Best,
Daniele

[niloufar]

Dear Mr.Varsano,
Thank you a lot. I will take these changes into consideration!