changes between 2.1 and 2.3/2.4-trunk inside calculate_dispersion_pairpot subroutine

[Marcin Zielinski]

Hi,

I've recently came across a problem with the 'calculate_dispersion_pairpot' 
subroutine inside the qs_dispersion_pairpot.F file. The 2.4 version of that 
subroutine is being executed ~6x slower than that of 2.1 (~6s in place of 
~1s). This was initially reported to me by one of our users running his 
jobs on our cluster. The builds of both 2.4 and 2.1 are practically the 
same.

Any thoughts and ideas about it?

regards and thanks in advance,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20130501/572a1fa7/attachment.htm>