[CPMD-list] discrepant total energies

[Kozin, I (Igor)]

Hello,
I was recently pointed out to the difference in 
the initial total energies computed using different
compilers. Particularly for cpmd-test/std/wat32/inp-1
IBM           -539.79205456 A.U.  (as provided)
Intel         -539.79205456 A.U.
PathScale     -539.79205494 A.U.
PGI           -534.04296070 A.U.

There is not much difference using Intel or PGI
on 32- or 64-bit platforms or using various compiler
options. Fortunately everything converges nicely to 
-551.05084246 A.U.
But still the question is why the initial PGI value
is so different?

Many thanks in advance,
Igor

I. Kozin  (i.kozin at dl.ac.uk)
CCLRC Daresbury Laboratory
tel: 01925 603308
http://www.cse.clrc.ac.uk/disco