On 10/31/18 11:06 AM, Ramon Guixà wrote:
Ok, I mean, It makes sense what you say. I still I don't understand though how this has not been spotted before, I mean, I presume this combination is widely used (CHARMM+GROMACS+GPUs) and I have replicated the problem using versions 5, 2016, and 2018...? I know I sound a little desperate, but is there any workaround I could use to proceed until this is resolved?
We've never had a problem with such a combination in any of our runs, but we're running on different GPU (P100 and V100) with ECC, newer compilers, etc. But we've had stable, microsecond runs using the CHARMM FF for membrane proteins without any issues. It could be something to do with your setup (you've got a very old gcc but recent CUDA, no ECC on any of your GPU, though in principle that shouldn't be the issue). Or maybe there's something going on with the software (bug) but to fix that, the developers need a reproducible, fast test case to dig into it. Running for 400 ns and hoping to see the same is a bit too much to ask :)
-Justin -- ================================================== Justin A. Lemkul, Ph.D. Assistant Professor Virginia Tech Department of Biochemistry 303 Engel Hall 340 West Campus Dr. Blacksburg, VA 24061 [email protected] | (540) 231-3129 http://www.thelemkullab.com ================================================== -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to [email protected].
