This had been a bugging problem for quite a while - not being able to run some codes in a parallel fashion. I knew how to compile MPICH with PGI compilers and also knew compiling VASP with MPI, but what I did not know (or was not sure about) was the following : Will a newly compiled version of MPICH conflict the one installed by default ? If so, how to get around it ?? All these doubts were clarified by a gentleman from Pacific Northwest Labs, and today I reinstalled/recompiled MPICH2-1.0, then compiled VASP4.6.26 (32 bit), ran some test jobs and updated the documentation. Just when I thought most of the obstacles related to Rama were done, news ones seem to be emerging : users not compressing their data and not reading MOTD flashed on their screens, leading to 20 percent disk usage in under 2 months…