> under the exact same parallelization and input Can you reduce the number of CPUs in the projection process? It may be worth trying the calculation sequence ’projwfc.x => bands.x’ with half the number of CPUs, and also try the calculation with a single CPU.
Are you using the oneAPI compiler? If so, can you set the environmental variable ‘FI_PROVIDER’ as follows in the projection computation? export FI_PROVIDER=psm3 Sincerely yours, 西館数芽 Kazume NISHIDATE Ph.D Department of Systems Innovation Engineering, Graduate School of Science and Engineering, Iwate University 4-3-5 Ueda, Morioka, Iwate, 020-8551 JAPAN Phone:+81-19-621-6391 [email protected], [email protected] https://sites.google.com/site/nisidatelab/ _______________________________________________________________________________ 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 [email protected] https://lists.quantum-espresso.org/mailman/listinfo/users
