Re: [AMBER] pmemd.cuda error with group restraint

From: Ross Walker <ross.rosswalker.co.uk>
Date: Fri, 1 Oct 2010 16:59:30 -0700

Hi Bongkeun,

I have tried this with AMBER 11 pmemd which works, and AMBER 11 pmemd.cuda
which gives the segfault you describe. I also tried this with the current
development tree version of pmemd.cuda and it works fine. The output files
are attached.

A patch to update the AMBER 11 version of pmemd which will fix this bug
(and provide parallel GPU support and additional fixes and tweaks) will be
coming very shortly (once I am done with a bunch of pointless reports) so if
you could hold on for a few more days that would be great.

All the best
Ross

> -----Original Message-----
> From: Bongkeun Kim [mailto:bkim.chem.ucsb.edu]
> Sent: Thursday, September 30, 2010 8:38 PM
> To: Ross Walker
> Subject: RE: pmemd.cuda error with group restraint
>
> Hi Ross,
>
> Yes, I've tried to uses pmemd.cuda first and checked it worked with
> sander.MPI.
> I can send all the file you need.
> I just run it with this command.
>
> pmemd.cuda -O -i equil.in -p samch3w.prmtop -c samch3w_heat1.rst -r
> samch3w_equil1.rst -o samch3w_equil1.out -x samch3w_equil1.mdcrd
> -ref samch3w_h
> eat1.rst
>
> I hope this could be solved soon.
> Thank you.
> Bongkeun Kim
>
> Quoting Ross Walker <ross.rosswalker.co.uk>:
>
> > Hi Bongkeun
> >
> > This is a known problem. It should be fixed very shortly with the same
> patch
> > that will enable parallel GPU runs.
> >
> > In the meantime if you send me your prmtop, inpcrd and mdin file I will
run
> > it with the development tree and check it works. Don't send it to the
list
> > since messages above 2MB in size are not permitted.
> >
> > I trust you have tried this with the CPU version of PMEMD and verified
that
> > it is just a problem related to the GPU version.
> >
> > All the best
> > Ross
> >
> >> -----Original Message-----
> >> From: Bongkeun Kim [mailto:bkim.chem.ucsb.edu]
> >> Sent: Thursday, September 30, 2010 4:46 PM
> >> To: amber.ambermd.org
> >> Subject: pmemd.cuda error with group restraint
> >>
> >> Hello,
> >>
> >> I built a self assembled monolayer surface with explicit water.
> >> Its structure is dialkane sulfide.
> >> In MD simulation, I wanted to fix the position of the center sulfur
> >> atom in each monomer by using following input file.
> >>
> >> equil.in---------------------------------------
> >> sam water: 100ps MD
> >> &cntrl
> >> imin = 0, irest = 1, ntx = 5,
> >> ntb = 2, pres0 = 1.0, ntp = 1,
> >> taup = 2.0,
> >> cut = 10.0, ntr = 1,
> >> ntc = 2, ntf = 2,
> >> tempi = 300.0, temp0 = 300.0,
> >> ntt = 3, gamma_ln = 1.0,
> >> nstlim = 50000, dt = 0.002,
> >> ntpr = 1000, ntwx = 1000, ntwr = 1000
> >> /
> >> Fix S atoms
> >> 10.0
> >> FIND
> >> S * * LIC
> >> SEARCH
> >> RES 1 225
> >> END
> >> END
> >> -----------------------------------------------
> >>
> >> With sander or sander.MPI, it does work with selecting 225 sulfur
atoms.
> >> -output file--------------------------------
> >> ----- READING GROUP 1; TITLE:
> >> Fix S atoms
> >>
> >> GROUP 1 HAS HARMONIC CONSTRAINTS 10.00000
> >> ALL ATOMS THAT MEET 1 OF THE FOLLOWING SPECIFICATIONS WILL BE
> >> INCLUDED IN GROUP BE
> >> LOW
> >>
> >> GRAPH NAME = S SYMBOL = * TREE SYMBOL = *
> >> RESIDUE TYPE = LIC
> >>
> >> GRP 1 RES 1 TO 225
> >> Number of atoms in this group = 225
> >> ----- END OF GROUP READ -----
> >> ------------------------------------------------------
> >>
> >> But if I run this with pmemd.cuda, I always have just a segmentation
> >> fault error.
> >> The command is
> >> --------------------------------------------------
> >> pmemd.cuda -O -i equil.in -p samch3w.prmtop -c samch3w_heat1.rst -r
> >> samch3w_equil1.rst -
> >> o samch3w_equil1.out -x samch3w_equil1.mdcrd -ref samch3w_heat1.rst
> -
> >> gpu 1
> >> ----------------------------------------------------------------
> >> In the input file, if the first argument is *, then pmemd.cuda works
OK.
> >> But the problem is the first one is assigned to any letter.
> >> --------------->error
> >> FIND
> >> S * * LIC
> >> SEARCH
> >> ---------------> error
> >>
> >> --------------->works fine
> >> FIND
> >> * * * LIC
> >> SEARCH
> >> ---------------->works fine
> >>
> >> Do you have any idea why pmemd.cuda fails with this FIND SELECT
> >> statement?
> >> Thank you.
> >> Bongkeun Kim
> >>
> >
> >
> >
>
>



_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

Received on Fri Oct 01 2010 - 17:30:03 PDT
Custom Search