[QE-users] Hybrid functionals using GPU accelerated QE

[Giovani Rech]

Dear QE community,

I’ve been testing the GPU accelerated version of QE using a hybrid functional (pbe0 and vdw-df-c090, to be exact, pun intended) and noticed that the use of the GPUs are virtually zero while the Hartree-Fock part is being calculated. This brings me to conclude that the the exact exchange calculation is not yet implemented for GPU. Can anyone confirm this info? I’ve not been able to find this information in the readmes or release notes.

Thank you,

Giovani Rech
Universidade de Caxias do Sul
Caxias do Sul - RS, Brasil