Dear Prof Case,
1. What happens when you set ntpr=1 on the GPU? Does the error occur
before the first energy is printed?
When I set ntpr=1, the error still appears on the first step.
1. We are certainly going to need a lot more information before anyone on
the list can offer real help. What version of Amber is involved? How
many atoms in the system? Posting your mdin file sometimes helps. You
mention "NVT": was there some reason for that? Did earlier runs on the
same system proceed without error?
Ans: I am using AMBER18 for my TI project.
Number of atoms in system= 48037 ( consists of my abl-kinase protein+ small molecule+water TIP3P+ions for neutralization)
My input file is
&cntrl
imin = 0, nstlim = 500, dt = 0.001,
irest = 1, ntx = 5, ig = -1,
tempi = 300.0, temp0 = 300.0,
ntc = 2, ntf = 1, tol = 0.00001,
ntwx = 10, ntwe = 10, ntwr = 10, ntpr = 10,
cut = 12.0,
ntt =3, ntb = 1, ntp = 0, gamma_ln=2.5,
nscm = 0,
ntr=1, restraintmask=':1', restraint_wt = 20.00,
icfe = 1, clambda = 0.31608, ifsc=1,
timask1=':C3N',timask2='',
scmask1=':C3N', scmask2=''
/
I am trying a NVT production (1ns) on all lambda windows printed here: But I get errors for 0.0080, 0.04794 ( Not other!?? Cant understand why)
0.0080
0.00922
0.04794
0.11505
0.20634
0.31608
0.43738
0.56262
0.68392
0.79366
0.88495
0.95206
0.99078
0.997
0.998
No specific reason why I did NVT but I was trying to simply replicate the publication based on AMBER-TI on gpus and they have done the production in NVT ensemble.
Publication is “
https://pubs.acs.org/doi/abs/10.1021/acs.jcim.9b00105”
Any help/suggestions/advice will be very very useful.
I am stuck on this last 3-4 months.
Thanks
Regards
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
From: David A Case<mailto:david.case.rutgers.edu>
Sent: 13 March 2020 16:54
To: AMBER Mailing List<mailto:amber.ambermd.org>
Subject: Re: [AMBER] TI runs on GPU
On Thu, Mar 12, 2020, Debarati DasGupta wrote:
>I don’t know why but doing NVT productions on GPU gives me this error.. I
>have tried 5 times and each time I get the same error.
>
>gpuNum=
>Error: an illegal memory access was encountered launching kernel kNLSkinTest
What happens when you run a short simulation with the same inputs
(probably just reducing nstlim and ntpr) on the CPU?
What happens when you set ntpr=1 on the GPU? Does the error occur
before the first energy is printed?
We are certainly going to need a lot more information before anyone on
the list can offer real help. What version of Amber is involved? How
many atoms in the system? Posting your mdin file sometimes helps. You
mention "NVT": was there some reason for that? Did earlier runs on the
same system proceed without error?
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Mar 15 2020 - 07:00:04 PDT
