The number of H\psi products, as well as the number of scf iterations, is not directly related to the size of the systems, but larger systems are often more difficult to converge than smaller ones, and there may be significant differences between "nasty" and "easy" systems.
Paolo On Tue, Nov 8, 2016 at 4:07 PM, Pablo García Risueño < [email protected]> wrote: > Thank you very much again, this help is very useful. In this point I > wonder if the (*N*h := number of *H*[image: $ \psi$] products needed by > iterative diagonalization) depends on the size of the system, or it is more > or less constant for different kinds of systems. > > Best regards > > 2016-11-08 12:58 GMT+01:00 Paolo Giannozzi <[email protected]>: > >> There is no similar page for the phonon code, but there is this remark: >> "*For phonon calculations, each of the 3Nat modes requires a time of the >> same order of magnitude of self-consistent calculation in the same system >> (possibly times a small multiple)." *More accurate estimates depend a >> lot upon details of the calculation that are not easy to predict (such as >> how quickly the system converges) >> >> Paolo >> >> On Tue, Nov 8, 2016 at 12:18 PM, Pablo García Risueño < >> [email protected]> wrote: >> >>> Thank you very much, prof. Giannozzi, that is what I was looking for. Do >>> you know if a similar page exists for the ph.x code? >>> >>> Best regards >>> >>> 2016-11-08 11:38 GMT+01:00 Paolo Giannozzi <[email protected]>: >>> >>>> You can find something here: >>>> http://www.quantum-espresso.org/wp-content/uploads/Doc/pw_us >>>> er_guide/node14.html >>>> >>>> Paolo >>>> >>>> On Tue, Nov 8, 2016 at 11:29 AM, Pablo García Risueño < >>>> [email protected]> wrote: >>>> >>>>> Dear Federico >>>>> >>>>> Thank you very much for your message. Unfortunately these tests seem >>>>> to measure the computing time as a function of the number of cores, and >>>>> not >>>>> as a function of the number of electrons//plane waves//etc. >>>>> >>>>> >>>>> >>>>> 2016-11-08 11:09 GMT+01:00 Federico Iori <[email protected]>: >>>>> >>>>>> Hi Pablo. >>>>>> >>>>>> Did you have a look at computational centers benchmark test? >>>>>> >>>>>> I can suggest you >>>>>> http://www.quantum-espresso.org/benchmarks/ >>>>>> >>>>>> www.hpc.cineca.it/content/quantumespresso-benchmark >>>>>> >>>>>> I am not sure they can provide all the info you are looking for >>>>>> regarding scaling vs all the parameters of th system thou, but maybe they >>>>>> can help somehow. >>>>>> >>>>>> [image: Université Paris-Sud] >>>>>> *Federico IORI* >>>>>> Marie Curie Fellow >>>>>> Laboratoire de Physique des Solides >>>>>> Bâtiment 510 - Rue André Rivière >>>>>> 91400 Orsay >>>>>> >>>>>> >>>>>> ------------------------------ >>>>>> *De: *"Pablo García Risueño" <[email protected]> >>>>>> *Para: *"PWSCF Forum" <[email protected]> >>>>>> *Enviados: *Martes, 8 de Noviembre 2016 10:50:10 >>>>>> *Asunto: *[Pw_forum] Question on scaling of Quantum Espresso >>>>>> >>>>>> >>>>>> Dear members of the Espresso community >>>>>> >>>>>> I would like to know how the total execution time of Quantum Espresso >>>>>> (pw.x and ph.x) scales with the different parameters of the system, like >>>>>> number of electrons, size of the box, ecutwfc, etc. It would be very >>>>>> useful >>>>>> for me to know in which parameters the scaling is linear, quadratic or >>>>>> cubic. Of course I can do tests on my own, but they would be just partly >>>>>> reliable due to Amdahl's issues. >>>>>> >>>>>> Thank you very much for your help. Best regards. >>>>>> >>>>>> -- >>>>>> -- >>>>>> >>>>>> Dr. Pablo García Risueño >>>>>> >>>>>> Institut für Physikalische Chemie, Universität Hamburg, Grindelallee >>>>>> 117, 20146 Hamburg >>>>>> >>>>>> Tel. +49 040 42 83 84 82 7 >>>>>> >>>>>> _______________________________________________ >>>>>> Pw_forum mailing list >>>>>> [email protected] >>>>>> http://pwscf.org/mailman/listinfo/pw_forum >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> Pw_forum mailing list >>>>>> [email protected] >>>>>> http://pwscf.org/mailman/listinfo/pw_forum >>>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> -- >>>>> >>>>> Dr. Pablo García Risueño >>>>> >>>>> Institut für Physikalische Chemie, Universität Hamburg, Grindelallee >>>>> 117, 20146 Hamburg >>>>> >>>>> Tel. +49 040 42 83 84 82 7 >>>>> >>>>> _______________________________________________ >>>>> Pw_forum mailing list >>>>> [email protected] >>>>> http://pwscf.org/mailman/listinfo/pw_forum >>>>> >>>> >>>> >>>> >>>> -- >>>> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, >>>> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy >>>> Phone +39-0432-558216, fax +39-0432-558222 >>>> >>>> >>>> _______________________________________________ >>>> Pw_forum mailing list >>>> [email protected] >>>> http://pwscf.org/mailman/listinfo/pw_forum >>>> >>> >>> >>> >>> -- >>> -- >>> >>> Dr. Pablo García Risueño >>> >>> Institut für Physikalische Chemie, Universität Hamburg, Grindelallee >>> 117, 20146 Hamburg >>> >>> Tel. +49 040 42 83 84 82 7 >>> >>> _______________________________________________ >>> Pw_forum mailing list >>> [email protected] >>> http://pwscf.org/mailman/listinfo/pw_forum >>> >> >> >> >> -- >> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, >> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy >> Phone +39-0432-558216, fax +39-0432-558222 >> >> >> _______________________________________________ >> Pw_forum mailing list >> [email protected] >> http://pwscf.org/mailman/listinfo/pw_forum >> > > > > -- > -- > > Dr. Pablo García Risueño > > Institut für Physikalische Chemie, Universität Hamburg, Grindelallee 117, > 20146 Hamburg > > Tel. +49 040 42 83 84 82 7 > > _______________________________________________ > Pw_forum mailing list > [email protected] > http://pwscf.org/mailman/listinfo/pw_forum > -- Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
_______________________________________________ Pw_forum mailing list [email protected] http://pwscf.org/mailman/listinfo/pw_forum
