Hi all,
Sorry I need to re-post this. It has been 2 weeks since my last e-mail
regarding to kernal error. But didn't find solution yet.
My problem is, I cannot run pmemd.cuda where it gives me an error:
Error: invalid argument launching kernel kNLGenerateSpatialHash
I managed to run pmemd in CPU, but not pmemd.cuda.
I already installed the latest NVIDIA driver with updated amber tools 13.
So please if anybody can help me on this.
Thank you in advance.
Regards,
Asfarina
---------- Forwarded message ----------
From: ASFARINA AMIR HASSAN <asfarina.siswa.um.edu.my>
Date: Mon, Apr 28, 2014 at 1:51 PM
Subject: Re: [AMBER] Error: invalid argument launching kernel
kNLGenerateSpatialHash
To: AMBER Mailing List <amber.ambermd.org>
Yea I already update with amber tools 13 as what you said.
I also try to run pmemd in CPU and it works!
But not for pmemd.cuda.
The "Error: invalid argument launching kernel kNLGenerateSpatialHash" still
occur.
Any other alternative that I can try?
On Thu, Apr 24, 2014 at 7:26 PM, Jason Swails <jason.swails.gmail.com>wrote:
> On Thu, 2014-04-24 at 17:01 +0800, ASFARINA AMIR HASSAN wrote:
> > Dear Amber users,
> >
> > I got this error when running pmemd.cuda:
> >
> > Error: invalid argument launching kernel kNLGenerateSpatialHash
> >
> > I follow as what Ross Walker mention in
> > http://archive.ambermd.org/201309/0011.html
> >
> > In my case, I do this:
> >
> > tar xvjf AmberTools12.tar.bz2
> > tar xvjf Amber12.tar.bz2
> > cd amber12
> > export AMBERHOME=`pwd`
> > ./update_amber.py --update
> > ./update_amber.py --update
> >
> > ./configure -cuda gnu
> > make install
> > make test
> >
> > But then the error still occurred.
> >
> > FYI, I'm using cuda 5.5, Ubuntu 12.04 with NVIDIA driver 331.67.
> > My graphic card is Quadro FX 3800.
> >
> > Please guide me on this.
>
> First, make sure you are using AmberTools 13. You can upgrade in-place
> using the command:
>
> cd $AMBERHOME
> ./update_amber --upgrade
> ./update_amber --update
> ./configure -cuda gnu
> make install
>
> That said, if the error is _still_ occurring, the first step in
> debugging pmemd.cuda issues is _always_ to try your simulation with the
> CPU versions of pmemd and see if it works there. If it does not work
> with the CPU version, then there is likely something wrong with your
> system and the error messages from pmemd or pmemd.MPI are often a bit
> more helpful than the corresponding messages from pmemd.cuda.
>
> HTH,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
>
> _______________________________________________
> 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 Thu May 08 2014 - 02:00:03 PDT
