FEP calculation with flat bottom restraints - Redmine #2236
Archive from user: Vikas Dubey
Simulation crashes on multiple nodes while running FEP calculations with flat-bottom restraints on ions of my system. I use gromacs 5.1.2. I have tried simulation on my PC as well. Commands and results are shown below :
with FB restraints
————————————————————————
gmx mdrun -deffnm md_17 -v -nt 24 : Runs fine
gmx mdrun -deffnm md_17 -v -nt 30 : Runs fine
gmx mdrun -deffnm md_17 -v -nt 32 : Runs fine
gmx mdrun -deffnm md_17 -v -nt 36 : Crashes
gmx mdrun -deffnm md_17 -v -nt 32 : Crashes
without_FB restraints (with normal restraints)
—————————————————————————-
gmx mdrun -deffnm md_17 -v -nt 24 : Runs fine
gmx mdrun -deffnm md_17 -v -nt 30 : Runs fine
gmx mdrun -deffnm md_17 -v -nt 32 : Runs fine
gmx mdrun -deffnm md_17 -v -nt 36 : Runs fine
gmx mdrun -deffnm md_17 -v -nt 48 : Runs fine
(from redmine: issue id 2236, created on 2017-08-24 by gmxdefault, closed on 2017-12-11)
- Relations:
- duplicates #2095 (closed)
- Changesets:
- Revision 62827ffd by Mark Abraham on 2017-08-25T16:18:27Z:
Fix flat-bottomed position restraints with multiple ranks
Reallocation was never done for flat-bottomed restraints,
so the indexing could go out of range, leading to segfaults.
Fixes #2236
Change-Id: I866f96684fc5a2fef6391ed62a70abdaa1581a33
- Uploads:
- eq_FB.tpr with flat-bottom restraints
- eq_normal_restraint.tpr with normal restraints