From: Haleh Abdizadeh (halehabdi_at_sabanciuniv.edu)
Date: Tue Aug 27 2013 - 05:13:11 CDT
I am using a C2050 GPU on a machine with 64GB RAM. As I checked the GPU
card has 64KB constant memory so there must be not a problem with the card.
Using a newer version of the NAMD gives an error implying that there is a
problem with conf file : ERROR(S) IN THE CONFIGURATION FILE
But the same conf file works fine without a CUDA card.
On Tue, Aug 27, 2013 at 12:15 PM, Axel Kohlmeyer <akohlmey_at_gmail.com> wrote:
> On Tue, Aug 27, 2013 at 11:06 AM, Haleh Abdizadeh
> <halehabdi_at_sabanciuniv.edu> wrote:
> > Hello everybody,
> > I am trying to run a namd simulation on a 80000 atoms system including a
> > huge protein, water molecules and ions on a linux server using ++local
> > /usr/local/NAMD_2.7_Linux-x86_64-CUDA/namd2 +idlepoll +devices 0
> *.conf >
> > *. txt using varying number of CPUs. When I try to use CUDA I get the
> > following message :
> > CUDA error memcpy to exclusions: invalid argument
> > the error discription in output file is as follows :
> > 23872 bytes of CUDA constant memory needed for exclusions
> > So I tried my job on different machines with maximum 3029592 B memory but
> > this time I get the error :
> > ERROR(S) IN THE CONFIGURATION FILE
> > the error in output file asks to specify cutoff distance and exclude.
> > Although my conf file is completely correct and I can run it when I do
> > use CUDA.
> > Could anyone please help me with this problem?
> for starters, don't ask for help for a way outdated NAMD executable.
> upgrade to a newer version and try again.
> second, what kind of GPU do you have? perhaps it is not capable of
> handling your system?
> > Thank you.
> > Best,
> > Haleh
> Dr. Axel Kohlmeyer akohlmey_at_gmail.com http://goo.gl/1wk0
> International Centre for Theoretical Physics, Trieste. Italy.
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:37 CST