Sorry for (possibly needlessly) bumping the thread, but I'd still be extremely grateful for any comments on my situation, since I really need the vdW-DF functionality that the developers were so kind to implement in SIESTA :)
2009/12/30 Vasilii Artyukhov <[email protected]> > Sorry, 320, of course. > > I've attached the files. The output is for 2 CPUs, the same thing happens > with more CPUs in trunk-320, while 3.0-beta works fine on 24 CPUs with this > input file. > > Oh yes, and the version of ifort and MKL is 10.something. > > Thanks for the quick response! > > 2009/12/30 <[email protected]> > > Dear Vasilii: >> >> Do you mean siesta-trunk-320 ? >> >> If so, please let us know more details (possibly input and output >> files and arch.make) so we can try to locate the problem. >> >> If you really meant trunk-302, try with siesta trunk-320 may be the >> issue has been already solved there. >> >> Yours, >> Jose A. Torres >> >> >> >> Quoting Vasilii Artyukhov <[email protected]>: >> >> Dear developers, >>> >>> I'm trying to migrate to the siesta-trunk-302 version of the package, and >>> I've run into some previously unseen trouble with a rather simple system: >>> a >>> graphane layer (in the XY plane) with an adsorbed molecule, using PBE & >>> all >>> the same stuff I've successfully been using for years now. The code >>> crashes >>> saying: >>> >>> odeMeshBox ERROR: undefined mesh distribution >>> >>> >>> The same input file seems to work perfectly fine with 3.0-beta, and >>> there's >>> no such error message when running in serial. I tried setting ProcessorY >>> to >>> different values, but without any success. What's the difference between >>> the >>> trunk version and the beta in the mesh decomposition, and how to handle >>> the >>> new code properly? >>> >>> Thanks in advance, >>> Vasilii >>> >>> >> >> >> -- >> Jose A. Torres, Ph.D. >> Manager of the SIESTA Software >> Tel: +34 914973805 >> >> >> >
