To run the job in parallel : ( for nmpi version)
Running Vasp-nmpi's

Making the pmg cluster familiar...
master is called =>  pmg.bose.res.in
[roby@pmg ~]$ hostname
pmg.bose.res.in

and we have 6  slaves
pmnode0 => pmg1.bose.res.in
pmnode1 => pmnode1.local
pmnode2 => pmnode2.local
pmnode3 => pmnode3.local
pmnode4 => pmnode4.local
pmnode5 => pmnode5.local

Setting up the mpd's
Log into the master and type the below commands:
[roby@pmg ~]$ which mpd
/opt/nmpiC/bin/mpd

If it shows the above then its fine else one should give the full path while using the mpd's

roby@pmg$ mpdallexit   
That will kill all the mpds running at present

Now type the below commands to set the mpd
roby@pmg$ mpd --daemon
roby@pmg$ mpdtrace -l


  (it will give  a pid number say 32858.. note that number)

Now log into the nodes which all u need to use and then type the command:
mpd -h pmg -p 32858 -n -d
earlier it was
mpd --host pmg --port 32858 --noconsole --daemon


Suppose you have done in all the nodes then when u r back in the master and type the command mpdtrace -l it gives the below (numbers are arbitary)
[roby@pmg ~]$ mpdtrace -l
pmg.bose.res.in_33211
pmnode5.local_33483
pmnode4.local_33494
pmnode3.local_33492
pmg1.bose.res.in_37314

[roby@pmg ~]$ mpdtrace
pmg
pmnode5
pmnode4
pmnode3
pmg1

Now all is set and now go to the folder where u need to run the job and make the host file ,,, u have to launch the job from the master...
So the host file is something like this...

[roby@pmg cdse_6th_sph_equill_6.184]$ cat host
pmnode0.local
pmnode0.local
pmnode0.local
pmnode0.local
pmnode0.local
pmnode0.local
pmnode0.local
pmnode0.local
pmnode3.local
.....
.... (like this u need to specify the processors )


Once this is done and suppose u have choosen 32 processors for the job then you need to type in the command from the master

[roby@pmg cdse_6th_sph_equill_6.184]$ mpiexec -machinefile host -np 32 /opt/vasp.4.6/vasp-nmpi-gama > LOG &

Happy Vasping :)
*********************************************************************************************************
Espresso in parallel

Now if u want to run the espresso in parallel u need to log into master and then type the command
in the directory where the files are there.

 [roby@pmg ~]$ mpiexec -machinefile host1 -np 8 /opt/espresso-3.2/bin/pw.x < SC.in > SC.out

where the host1 is a similar file as the one discussed above for the nmpi vasp.
***********************************************************************************************************