Dear AMBER community,
I've recently been encountering an error when I try to run a single-walker
flooding-type adaptively biased MD run using pmemd.cuda. The run will
proceed as expected until around 1/10 of the expected duration, at which
point it will quit and write out a single error message:
cudaMemcpy GpuBuffer::Download failed an illegal memory access was
encountered
This has occurred in both AMBER18 and AMBER20, and has occurred with
identical simulations that ran properly a few months ago. I'm wondering if
anyone has ever experienced this issue, and if so what was causing it? Any
thoughts would be appreciated. The relevant .cntl and .in files are below:
.cntl file:
&cntrl
imin=0, ntx=1, irest=0, ntxo=2, ntpr=25000,
ntwr=250000, ntwx=25000, ntwv=250000, ntwe=250000,
ntf=2, ntc=2, ntb=1, nstlim=25000000, dt=0.002,
ibelly=0,iwrap=1,ntr=0, nmropt=1,
temp0=300.0, tempi=300.0, tautp=1.0,
ntt=3, gamma_ln=1.0, ig=-1,
ntp=0, pres0=1.0, comp=44.6,
taup=0.2, tol=0.00001, infe=1
&end
&ewald
order=4,
ew_type=0
&end
/
&wt type='REST',value1=1.0,value2=1.0, /
&wt type='END' /
LISTOUT=src/NoS_Stem.restraint
DISANG=src/stem_base.rst
/
&abmd
mode='FLOODING'
monitor_file='bias/NoS_Dih.mon.1.txt'
monitor_freq=500
umbrella_file='bias/NoS_Dih.bias.nc'
snapshots_basename='bias/NoS_Dih.snapshots',
cv_file='src/cv_nstruct_dih.in'
timescale=1.0
/
&wt type='END'
/
.in file:
&colvar
cv_type = 'N_OF_STRUCTURES'
cv_ni = 29, cv_nr = 86,
cv_i = 156, 171, 187, 202, 218, 233, 249, 264, 280, 313, 627, 661, 1619,
1653, 0,
1059, 1075, 1091, 1106, 1120, 1136, 1152, 1167, 1182, 1197, 1518,
1551, 695, 728
cv_r = [coordinates, very long list]
cv_min=0.0,cv_max=2.0
resolution=0.05
/
&colvar
cv_type = 'COM_TORSION'
cv_ni = 12,
cv_i = 355, 553, 0, 256, 648, 0, 225, 1572, 0, 32, 1765, 0
cv_min = -3.15,cv_max = 3.15
resolution = 0.315
/
Sincerely,
Owen
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Feb 05 2023 - 16:00:02 PST