[Pw_forum] problems running test-system on Alpha. Any ideas?

[Andriy H. Nevidomskyy]

Dear PWSCF users,

I have trouble running the test example with pw.x (version 1.2.0).
This is part of the example No.1, which I downloaded from the PWSCF
web-site, and it's about self-consistent calculation of Cu-metal.
The run on DEC-Alpha fails with the following error message:

---------------------------------------------------------
 forrtl: error (65): floating invalid
 IOT trap (core dumped)
---------------------------------------------------------
OR, on slightly different machine (also Alpha):
----------------------------------------------------------
forrtl: error (75): floating point exception
IOT trap (core dumped)
----------------------------------------------------------

The analysis of the output file shows that the execution fails even before
the first SCF step was taken.
The input file that I used reads as follows:

               <cu.scf.in>
----------------------------------------------------------------
 &control
    calculation='scf'
    restart_mode='from_scratch',
    pseudo_dir = '/home/giannozz/O-sesame/pseudo//',
    outdir='/home/giannozz/tmp//'
    prefix='cu'
    tstress = .true.
    tprnfor = .true.
 /
 &system
    ibrav = 2, celldm(1) =6.73, nat= 1, ntyp= 1,
    ecutwfc = 25.0, ecutrho = 300.0
    occupations='smearing', smearing='methfessel-paxton', degauss=0.02
 /
 &electrons
    conv_thr = 1.0e-8
    mixing_beta = 0.7
 /
ATOMIC_SPECIES
 Cu 63.55 CuUS.RRKJ3.UPF
ATOMIC_POSITIONS
 Cu 0.0 0.0 0.0
K_POINTS (automatic)
 8 8 8 0 0 0
--------------------------------------------------------------------

I played with different values of the input params: tried increasing
_nbnd_, changing the _smearing_ scheme, increasing _degauss_ value - all
in vain, the error is steadily reproducible!

And of course, this error doesn't occur in the output file given by
the code developers as a reference to the test [reference says that test
was run on PC Pentium-4 machine].
Shall I thus assume, that this is a platform-dependent issue, specific
for Alphas?
However, the calculation for another metal, Al, doesn't produce this
error! Why does it happen for Cu then?


I would be very grateful for any insight that you could share with me on
this problem.

P.S> *It should be noted that presicely the same
     *error occurs in the SCF-calculation, as well as in the
     *band-structure calculations for another test example - Ni.

Thanks!
Andriy Nevidomskyy

.------------------------------------------------------------------.
|                        Mr Andriy H. Nevidomskyy                  |
|                                                                  |
|     TCM Group, Cavendish Laboratory, University of Cambridge     |
|          Madingley Road, Cambridge CB3 0HE United Kingdom        |
| Tel:+44-1223-337433,Fax:+44-1223-337356, E-mail: ahn22 at cam.ac.uk |
`------------------------------------------------------------------'