So far I have been successfully using Dalton 2015. However I needed Dalton compiled with some changes in mxcent.h and qm3.h files so I compiled Dalton 2016.0.
Now, the issue I run into is related to memory allocation when running Dalton in parallel (see the attached files and especially the DALTON-batch-system.out one).
I call Dalton with
and total memory for the job is 40 000 MB. This actually does not matter because I run into exactly the same issues if I use
Code: Select all
dalton -N 6 -mb 6000 DALTON.dal MOLECULE.mol
-mb 15000 and 115 000 MB of total memory. What is more the same job in Dalton 2015 took much less memory.
On the other hand, when I run dalton sequential there is no problem with memory allocation.
I have used both intel-15.0 and intel-13.1 to compile Dalton 2016, with the same results. Is this specific to my OpenMPI version?
PS I have changed MXQM3 to 15000 in qm3.h and set MXCENT_QM = 1200 MXCENT = 15000 in mxcent.h