Hi all,
I am trying to run an explicit solvent constant pH simulation with
pmemd.cuda which constantly fails after a few ns giving the following error:
“Error: an illegal memory access was encountered launching kernel
kClearForces
cudaFree GpuBuffer::Deallocate failed an illegal memory access was
encountered”
I already tried reducing ntpr and ntwx to look for anything pointing to the
problem, but couldn’t find anything. The system is a protein and approx.
26000 atoms.
I already started a run using pmemd.MPI to get a more verbose error message.
I am titrating all HIP, AS4, GL4 residues, giving a total of 24.
I’d really appreciate any advice to identify the problem.
My input file:
production phase, NVT, 250 ns
&cntrl
imin = 0,
ntx = 5,
irest = 1,
ntpr = 500,
ntwr = 625000,
ntwe = 2500,
ntwx = 500,
ntwprt = 2786,
ioutfm = 1,
iwrap = 1,
cut = 9.0,
ntt = 3,
ntb=1,
ntp=0
temp0 = 300,
gamma_ln = 1.0,
ig = -1,
nstlim = 125000000,
dt = 0.002,
icnstph=2, ntcnstph=100, ntrelax=200, solvph=5.0, saltcon=0.001, ntc=2,
ntf=2
/
Andreas Tosstorff, M.Sc. Chemistry, PhD Student
Ludwig-Maximilians-University Munich
Department of Pharmacy
Pharmaceutical Technology & Biopharmaceutics
Butenandtstr. 5
D-81377 München
Tel: +49 89 2180 77059
Fax: +49 89 2180 77020
<mailto:Andreas.Tosstorff.cup.uni-muenchen.de>
Andreas.Tosstorff.cup.uni-muenchen.de
<
http://www.cup.uni-muenchen.de/pb/aks/winter/>
http://www.cup.uni-muenchen.de/pb/aks/winter/
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Oct 08 2018 - 13:30:03 PDT
