[Pw_forum] parallelization of phonon calculations

[Paolo Giannozzi]

On Jul 25, 2012, at 16:13 , William Parker wrote:

> Each image will output to its own output file labeled out.i_0 where  
> i runs from 1 to N_image - 1.  The frequencies in these files will  
> be garbage unless all of the representations for a particular q- 
> point are contained in one image.  After you run the image- 
> parallelized run, you must copy the data-file.*.*.xml files (not  
> the data-file.*.xml files) to the _ph0/(prefix).phsave/ directory  
> and run again on one image with recover=.true. in your input file.   
> This will produce the correct frequencies in the output file and  
> the (prefix).dyn* files.

funny: "image" parallelization (the one done with MPI, not with separate
calculations on a grid) should automatically perform the final step of
collecting results.

P.
---
Paolo Giannozzi, Dept of Chemistry&Physics&Environment,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222