[Pw_forum] forces in parallel espresso 3.0

[Axel Kohlmeyer]

On Tue, 7 Feb 2006, Andrea Floris wrote:



AF> Dear users,
AF> I am running in parallel a scf calculation,
AF> with the setup:
AF> 
AF> pwscf version: 3.0
AF> compiler: pgi 5.2-4

andrea,

can you please try with a different compiler.
pgi-5.x, especially the opteron variants are 
known to miscompile several parts in differnt
DFT codes. this may be one of the cases. 

does the crash also happen in the serial case?

ciao,
    axel.

AF> mpi: mpich2
AF> system: cluster OPTERON 64-bit
AF> 
AF> I want to report that within my setup, pw.x  crashes when it calculates
AF> the forces, (i.e. when including tprnfor = .true. in the namelist
AF> &control), producing
AF> the message:
AF> ----------------------
AF> .....
AF> rank 3 in job 1 n34_16977 caused
AF> collective abort of all ranks exit status of rank 3: killed by signal 11
AF> rank 2 in job 1 n34_16977 caused collective abort of all ranks
AF>   exit status of rank 2: killed by signal 11
AF> rank 1 in job 1  n34_16977   caused collective abort of all ranks
AF>   exit status of rank 1: killed by signal 11
AF> ------------------------
AF> 
AF> With previous versions this doesn't happen. And it doesn't happen also
AF> with version 3.0 running in serial...
AF> Has this happened in other setup?
AF> 
AF> many thanks
AF> ciao,
AF> Andrea
AF> _______________________________________________
AF> Pw_forum mailing list
AF> Pw_forum at pwscf.org
AF> http://www.democritos.it/mailman/listinfo/pw_forum
AF> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.