I have a bizzar problem. I did 21 simulation for 4 ns each. These were equ'd for 4ns prior, then taken from an assembly run over 4ns so between 300 pico and 4ns additional each. A reviewer wants a couple extnded ends added,say like 4 extra ns, which I can only do 1-4 at most due to time limits (they take several days with 4-5 nodes, 8cpu quad core amd) Now I just simply took the end run .mdp files, added these on, generated .tpr file, and a short check to generate a checkpoint .cpt. I ran one on the cluster here in bern, one on a home PC and submitted one to the dutch/euro grid system. The one on the grid in bern works fine, the one on my PC as well (intel 8 core), however it only reaches 75,000 of 20 million time steps necessary in 8 hours, I could never go beyonda 0.0005 time step (well 0.0008 but kept it at the former due to ease of changing file parmeters), any case, when I submit it to the Eu/dutch grid it crashes at at step 5000 with a PME, non-equilibrated error. My question is why does it run on one, the exact same tpr tested, but not the other?
I looked at versions 4.5.5 my pc, 4.5.7 at unibe and a altered version of 4.5.3 on the dutch/eu grid. Other than that, I can not find a problem, but woundered, as use of an extra que would have helped but most likely not now, as I have a response time. I can not fathom why it would give such large changes however in bonds/angles in one but not the other as the versions are not that different? Was there a large change between these versions, or other possibilities?
Stephan Watkins
-- gmx-users mailing list gmx-users@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-requ...@gromacs.org. * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
