[Pw_forum] problem with pp.x and dos and a question about vectors in ILDOS calculation
sclauzer at sissa.it
Tue Feb 15 22:54:54 CET 2011
1) buggy compiler? which versions of gfortran are you using? is it the same on both machines? Have you tried with a different compiler?
2) try to define better the configuration giving problem: which compiler, which QE version? If you compare different compilers/QE versions on the two machines it is more difficult to spot out where is the problem.
3) monitor memory usage (with top command for instance: you can execute it on the background and print to file snapshots of system usage)
4) try to understand at which point of execution the code stops. open the source files of the code that crashes, put some print instructions. Better, use a debugger...
Il giorno 15/feb/2011, alle ore 19.48, pari shok ha scritto:
> Dear All,
> problem with pp.x and dos and a question about vectors in ILDOS calculation.
> Firstly, the problem with pp.x and dos.x:
> I downloaded QE in two machines:
> intel (R) pentium (R) 4 cpu 3 GHz with total memory 2GB, uses gfortran and
> intel (R) core (TM) i7 cpu 870 @2.93 GHz, with 4GB total memory, uses gfortran.
> everything is working perfectly well with the first machine (with espresso-4.1.2), but dos.x and pp.x both crash with "out of memory/ memory allocation" error in the second machine for big structures (espresso-4.2.1 and espresso-4.2). It should be noted that pw.x works properly and the "memory allocation" error happens just for dos and post processing. Sometimes, it seems that both pp.x and dos.x are frozen. In this case, nothing is written to output file and the programs don't even send any memory messages.
> I downloaded ATLAS and modified "make.sys" accordingly but it didn't help. I used "ulimit -s unlimited" command, it didn't help either.
> I should emphasise that dos.x and pp.x in the old machine work properly for the same inputs. Therefore, the input files don't have any problems.
> I am really obsessed with it. I would really be grateful if anybody who had experienced the same problem could help me somehow.
> Secondly, a couple of questions about vectors in ILDOS:
> 1. If the origin of ILDOS calculation is set to (000) and the vector is set to (001), does the ILDOS calculations spans the whole slab in "z" direction?
> 2. Does ILDOS in one direction (along one vector) gives the LDOS on the plane perpendicular to that vector in each point?
> I really appreciate your help.
> This is the second time that I've subscribed for the forum in the hope that this time my subscription would be successful.
> P Shok
> PhD at UMD
> Pw_forum mailing list
> Pw_forum at pwscf.org
§ Gabriele Sclauzero, EPFL SB ITP CSEA
PH H2 462, Station 3, CH-1015 Lausanne
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the users