[gmx-users] Desktop Freezes While Using mdrun-gpu

[Szilárd Páll]

Hi Solomon,

Just stumbled upon your mail and I thought you could still use a
answer to your question.

First of all, as you've probably read on the Gromacs-GPU page, a) you
need a high-performance GPU to achieve good performance (in comparison
to the CPU) -- that's the reason for the strict compatibility list and
the warning about overriding this.

Ideally, the Gromacs-GPU MD code should be able to run on every
compute-capability-wise compatible GPU. However, it does not really
make sense to use mobile or other lower-end GPUs. Moreover, I have
seen freezes and extreme slowdowns when running on such hardware,
especially if the graphics card is also used for display.

Regarding your problem, I don't have an instant solution, but as I
said, except for trying it out you won't gain much with the 330M chip.
What OS are you using? Have you tried with the latest drivers and CUDA
library? If you are using Linux, try to stop the X server and start
the simulation in console mode.

Cheers,
--
Szilárd



On Sat, Dec 4, 2010 at 7:17 PM, Solomon Berman <smberman at bu.edu> wrote:
> Hello friends,
> I'm writing today to see if anyone could offer further insight/commentary
> than the current available documentation and comments from the program
> itself concerning the following problem.
> I am currently using mdrun-gpu, v 4.5.3, utilizing my GPU, an NVIDIA GeForce
> GT 330M.  When I run the program, my desktop freezes, though I am able to
> move my cursor about.  I can hear the fan growing louder and the computer
> heating up, as it always does when I use the CPU mdrun.
> Browsing the available documentation, I recognize that my particular GPU is
> not listed as one of the tested.  However, from reading further, and from
> the error codes generated by mdrun-gpu, I am suspect that it is possible to
> use a GPU not listed in the manual, so long as the other requirements (such
> as the use of the OpenMM library) are met.
> Could anyone who is using the GPU version perhaps offer some insight into
> what is occurring and any potential fixes that may be successful?  I would
> be most appreciative?  Also, if I haven't provided enough information, I'm
> more than happy to answer any questions quickly and succinctly to do so!
> Many thanks for any help anyone can provide!
> Best,
> Solomon Berman
> Department of Chemistry
> Boston University
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>