Dear Developers,
I am doing Steered Molecular Dynamics Simulations with AMBER 14 in both
explicit and implicit solvent and I am facing the following problems.
1) With implicit solvent, for a protein of 1689 atoms on 1 GPU - 1 CPU,
setting the non-polar term to zero, gbsa=0 (default), I got a
performance of around 300 ns/day. However, by setting gbsa=1 the
performance falls to 4 ns/day (which is even slower than the explicit
solvent simulations for protein + water with 34 Kb atoms, 90 ns/day also
in 1 GPU - 1 CPU). Is there any GPU/CPU combination for implicit
solvent, including the gbsa term, with a better performance in AMBER? Is
this issue being considered in the new releases of AMBER?
2) For the pulling simulations I am using the option "jar=1". With this
option the distance between two given atoms is changed at a constant
velocity, but I can not control the direction of the applied force (I
need one of the atoms being pulled to move along x-axis). I thought to
solve this problem by restraining the motion of the pulled atom to the
x-axis with the option "restraint_wt", but this option seems to apply
restrains in all the axis simultaneously. Is there any way to control
the direction of the pulling in AMBER or to restrain the motion of one
atom in yz while moving in the x direction?
Thanks in advance,
Best regards,
Ana
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Sep 22 2016 - 01:00:03 PDT
