[Q-e-developers] problem with pp

[Nicola Manini]

I confirm that Paolo's diagnosis is correct: the bug is caused by an 
interaction of gfortran (4.1.2 20080704) and iotk.
The bug is not reproduced when everything is recompiled (and re-run, since 
the binary outputs are incompatible) under ifort.
Incidentally, pw.x execution is almost 2 times faster under ifort... ;-)

Thank you all!
			Nick


On 01/11/2012 10:01 AM, Paolo Giannozzi wrote:
> Please choose one of the following:
>
> a) the data file ge100-p2x1-out.save, generated by v.4.3.2 of pw.x,
>     cannot be read by v.4.3.2 of pp.x, while the same procedure was
>     working in v.4.0
> b) the data file ge100-p2x1-out.save. generated by v.4.0 of pw.x,
>     cannot be read by v.4.3.2 of pp.x,
> c) none of the above (please specificy)
>
> I can anticipate that if a) applies, with 99.99% probability
> it is a (well-known) problem of miscompilation (by gfortran) of
> iotk; if b) applies, we need the pw.x data to see why it happens
>
> P.