Hello everyone, 


I’ve run into an issue while trying to run SCAN calculations, and I believe it 
might be related to compilation, although I’m not entirely certain. 

We have the following setups: 

    * CPU cluster, where we compiled libxc 6.2.2 with QE 7.3.1 and everything 
works fine. 
    * Larger GPU cluster with the NVIDIA versions of libxc 7.0.0, QE 7.2 and QE 
7.4 , and here the calculations fail (w both QE versions) for inputs that 
worked without issue on the CPU cluster. 

and on the GPU cluster we started getting the following warnings 

WARNING: libxc functional with ID 263 depends 
on external parameters: check the user_guide of 
QE if you need to modify them or to check their 
default values. 

WARNING: libxc functional with ID 263 is CAM, 
long range exx is not available yet (if needed). 
[w02] libxc functional with ID 263 is still 
in development. 

which on the CPU cluster don't show up. From what I've seen in the source code 
these warnings are triggered by some variable called libxc_flags, however I 
haven't been able to pinpoint the root of the issue. 

Could you help me troubleshoot this? Any comments on how to address these 
warnings or find any differences in the compilation setup between the CPU/GPU 
would be greatly appreciated. 

Best regards and thanks in advance, 
Sofía 

-- 
Lic. Sofía Anna Scozziero 

Comisión Nacional de Energía Atómica 
Gerencia de Reactores y Centrales Nucleares 

Oficina 15, 1er Piso, Edificio 3 
Centro Atómico Constituyentes 
Av. General Paz 1499, Villa Maipú CP1650 

Interno: 7227 
e-mail: sofiascozzi...@cnea.gob.ar 
_______________________________________________________________________________
The Quantum ESPRESSO Foundation stands in solidarity with all civilians 
worldwide who are victims of terrorism, military aggression, and indiscriminate 
warfare.
--------------------------------------------------------------------------------
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list users@lists.quantum-espresso.org
https://lists.quantum-espresso.org/mailman/listinfo/users

Reply via email to