From: Ashkan Shekaari (shekaari_at_email.kntu.ac.ir)
Date: Mon Jul 27 2020 - 07:03:54 CDT
Are you using gpu along with cpu hyperthreading?
If so, you will not have speed-up. Therefore,
0- Change to linux os.
1- Turn off cpu-hyperthreading.
2- Download and install the latest cuda version of namd, not the multicore one.
-- All the best, Ashkan Shekaari, Ph.D Candidate in Solid State Physics, K. N. Toosi University of Technology, Tehran, Iran From: "Asmi Mahmood" <asmi_mahmood_at_REMOVE_yahoo.com> To: "namd-l" <namd-l_at_ks.uiuc.edu> Sent: Monday, July 27, 2020 1:36:28 PM Subject: namd-l: NAMD on 24 threads Hi I need some help in optimizing my system efficiency for Simulation. Previously i was running NAMD on system with 4 processor on windows 10 and get 10 ns results in 10 days. Now as i have upgraded my system to dual processor each with 12 core and 12 threads (total 24 processor) with NVIDIA quadro fx 580 i was expecting a greater efficiency but still not getting desirable results as 10 ns is taking 5 days. It is to mention that i am using OS window 7 on my upgraded system and NAMD 2.13 multicore 64 bit.. what is reason of so low speed on 24 processors? In addition, i am planing to upgrade OS to linux and wants use CUDA environment if can be done on my system. Please guide me with this GPU card can i use CUDA as it supports CUDA 1.1 (as in its specifications). If yes which NAMD version should i use. Asma Tariq
This archive was generated by hypermail 2.1.6 : Thu Dec 31 2020 - 23:17:13 CST