Difference between revisions of "VT GPGPU Use Case moldyngrid UseCase1"
Revision as of 17:02, 19 June 2015
|Main||EGI.eu operations services||Support||Documentation||Tools||Activities||Performance||Technology||Catch-all Services||Resource Allocation||Security|
Life Sciences. Note well: also WeNMR has a GPGPU use case based on GROMACS. Local resources in Utrech are used.
Institute or Community
- IMBG of NAS of Ukraine
- EGI VO Community: moldyngrid
We are using Gromacs - ftp://ftp.gromacs.org/pub/gromacs/gromacs-4.5.5.tar.gz (free) and OpenMM (plugin for gpu) - https://simtk.org/home/openmm (free, but need registration) The GROMACS (mdrun) should be patched by OpenMM. There is tutor - http://www.gromacs.org/Documentation/Installation_Instructions/GPUs?highlight=GPU. For automatization we need ~12 steps before MD simulation. I have propose to make in the first approximation. For fast result (in our VT) we can do just only for last step (for mdrun). All others steps I can do it by myself (on local cluster). For this last step we need md.trp file (project file) to run with mdrun (file from gromacs). The problem is, what we should to do with "mdrun" file before submit MD job. Is it need to be compiled and send with project file or mdrun should be compiled on each clusters before I will send job. Its technical question for admins.