From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Sat May 12 2018 - 01:10:39 CDT
I am nearly always observing a slowdown of FEP. during execution on a
NextScale cluster. A considerable slowdown: in the last simulation, on a
single node (36cores), from 0.0025s/step to 0.0060s/step. As such a
degradation of performance occurs variably from the first window to later
windows, while the FEP has to be carried out entirely, it becomes
problematic setting up the calculation, also because I am trying to have as
many steps as possible with alchequil and FEP.
I wonder whether the same problem has been ever observed.
Thanks for sharing experience.
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:19:54 CST