Hi Stefan,
AMBER does not make use of any temporary disk space when running so there really aren't many advantages to having local disk on a node for AMBER unless you have a slow or overloaded cluster wide storage. One advantage of local disk on a node might be for writing output data there as part of a workflow that also did the analysis and then discarded the intermediate simulation data but that would likely be a very specific use case.
All the best
Ross
> On Feb 13, 2018, at 12:27, Lenz, Stefan <stefan.lenz.uleth.ca> wrote:
>
> Hello,
>
> I have noticed that on the suggested hardware listed on
> http://ambermd.org/gpus/recommended_hardware.htm that system storage is not
> listed for the rack mount nodes. I was wondering if pmemd (MPI, GPU or CPU
> version) makes use of the disk that may be present on CPU or GPU nodes on a
> cluster.
>
> Thanks for any help that you may be able to provide,
>
> Stefan Lenz
> _______________________________________________
> 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 Wed Feb 14 2018 - 07:00:02 PST
