Dear Jaroslaw Please consider that small size systems are only toys for early testing of methods. If you wish to perform real calculation time analysis, use larger systems available in Matpower: >1000 bus. Best regards
De : [email protected] [mailto:[email protected]] De la part de Jaroslaw Krata Envoyé : mercredi 30 novembre 2016 06:26 À : [email protected] Objet : RunPF vs GPU calculations Hello Matpower Team, I have a quick question related to runpf calculation time. Some time ago I started to use matpower to calculate ac power flow. Due to the fact that I need thousands of power flows I wish to run it effectively. I run Matlab Profiler and I have found out that there is a lot of computation time taken by simple algebra. I.e. (calculation time for 1000 iterations of runpf of 49 bus system): Function Calls Total time Self time runpf 1000 1.692 0.183 s newtonpf 1000 0.684 0.552 s If you have a closer look into newtonpf, the 50% of time is taken by "dx = -(J \ F);" operation! Similarly, dSbus_dV consumes >60% of time by only two lines: "dSbus_dVm = diagV * conj(Ybus * diagVnorm) + conj(diagIbus) * diagVnorm; dSbus_dVa = 1j * diagV * conj(diagIbus - Ybus * diagV);" These 3 lines take in total 0.420 out of 1.692 second. Have you considered to move this algebra to CUDA GPU? Are there any known constraints not to use CUDA? https://developer.nvidia.com/matlab-cuda https://www.youtube.com/watch?v=PwCAWxPcDkw BR, Jaroslaw "Ce message est destiné exclusivement aux personnes ou entités auxquelles il est adressé et peut contenir des informations privilégiées ou confidentielles. Si vous avez reçu ce document par erreur, merci de nous l'indiquer par retour, de ne pas le transmettre et de procéder à sa destruction. This message is solely intended for the use of the individual or entity to which it is addressed and may contain information that is privileged or confidential. If you have received this communication by error, please notify us immediately by electronic mail, do not disclose it and delete the original message."
