[Pw_forum] cvs version in parallel, a problem

Jess Kondor kondor.jess at gmail.com
Thu Dec 27 18:59:45 CET 2007


Dear developers,

 I understand, that CVS version is under development and might not work, but
in that version a bug in Berry Phase calculation is fixed, so I prefer to
use CVS version instead official one.

I have noticed that pwscf  crashes in parallel (compiled by ifort 9.1,
mvapich) , when more than 2 cores is used. Test example is an 'atom.in' file
provided in ' tests/' . The error message is:
.....
     Initial potential from superposition of free atoms

     starting charge    6.00000, renormalised to    6.00000

     negative rho (up, down):  0.101E-04 0.000E+00
     Starting wfc are    4 atomic +    2 random wfc
mpirun_rsh: Abort signaled from [1]

Checking atom...1 - MPI_REQUEST_FREE : Invalid mpi_request handle
[1] [] Aborting Program!
done.

Does anybody notice the same problems?

 Best wishes,

-- 
=====================================
Jess Kondor,
PhD Student

PICYT, Av. Venustiano Carranza 2425-A,
San Luis Potosí 78210, México
=====================================
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20071227/3b8b456a/attachment.html>


More information about the users mailing list