[Wien] Re: [Wien] .machines file for fine grained parallel executions

Jorissen Kevin Kevin.Jorissen at ua.ac.be
Wed Aug 20 10:52:36 CEST 2003


Hello Tom,
 
if the problem is in the lapw1_mpi - executable, then I can't be of any help.
 
Just a very small hint in case you want to check the behaviour of lapw1para:
Go into the file lapw1para (you will find it in your WIENROOT directory).  Somewhere in the beginning, a few parameters are defined.  One of them is 'debug' and is set to 0 by default.  Raise it to 3.  Then rerun lapw1_mpi, and post the output (which should be quite lengthy now) to the mailing list.
This 'debugging output' is usually sufficient to determine where things go wrong, if it is a problem in the script (and not in the program lapw1 itself).
 
 
 
Kevin.
 

	Tom : I also tried to call directly lapw0_mpi and lapw1_mpi by invoking following
	commands.
	
	#mpirun -np 3 -machinefile m /home/wien2k/WIEN2k_03_3/SRC_lapw1/
	lapw1_mpi lapw1.def
	
	 
	Kevin : I think you should use lapw1_1.def?
	 
	LOPW errors are usually said to indicate errors in your struct-file (if I remember correctly, lopw is a subroutine that defines the wave vectors of the plane waves used for the expansion of our wave functions?), but I'm not sure that it can't mean anything else in your mpi-calculation.
	 
	 
	 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 5086 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20030820/0c000733/attachment.bin


More information about the Wien mailing list