[molpro-user] Fwd: CCSD(T) running time

Quan Phung manhquank9 at gmail.com
Mon Jan 20 18:13:09 GMT 2014


Hi,


On Mon, Jan 20, 2014 at 6:08 PM, Jeff Hammond <jeff.science at gmail.com>
 wrote:

> The report of zero CPU utilization suggests you are not making
> progress on the computation.  You could be slamming on swap space or
> thrashing disk.  Or perhaps something has failed and the job is hung.
>
> Yes, I noticed that three processors are running (100%) and there is
swapping between processors, meaning something is running.



> The negative number is probably just an overflow because you're
> running for an unreasonably long time.  You should determine if your
> job is making progress using the MolPro log file(s) rather than top or
> equivalent.
>
> In ROHF-CCSD(T), how can you know if the calculation is making progress
during the (T) part?
As usual, the calculation just finish CCSD and then print the (T) part at
the end. There is no % report like in RHF-CCSD(T).


> As a side note, I ran ROHF-CCSD(T) with 39 electrons in 711 AOs in 72
> minutes last week (using a completely different setup altogether).
> You are far too patient and should look into getting a bigger
> computer.
>
My calculations consist of 90 electrons, open shell, C1 symmetry, so that
it will take a lot of time.


On Mon, Jan 20, 2014 at 6:08 PM, Jeff Hammond <jeff.science at gmail.com>wrote:

> The report of zero CPU utilization suggests you are not making
> progress on the computation.  You could be slamming on swap space or
> thrashing disk.  Or perhaps something has failed and the job is hung.
>
> The negative number is probably just an overflow because you're
> running for an unreasonably long time.  You should determine if your
> job is making progress using the MolPro log file(s) rather than top or
> equivalent.
>
> As a side note, I ran ROHF-CCSD(T) with 39 electrons in 711 AOs in 72
> minutes last week (using a completely different setup altogether).
> You are far too patient and should look into getting a bigger
> computer.
>
> Jeff
>
> On Mon, Jan 20, 2014 at 9:55 AM, Quan Phung <manhquank9 at gmail.com> wrote:
> > Dear Molpro users,
> >
> > Currently, I'm doing very large R-CCSD(T) calculations (with about 600
> basis
> > functions).
> > Based on my experience from other similar systems, I know that the
> > calculation will
> > take place at least 30 days. After running about 20 days, I noticed that
> the
> > system reported
> > the running time as a negative number -5965423h-14:-8, and it is just
> > hanging there (using htop or top). The processors are still running but
> I'm
> > not sure if the calculations continue or not.
> > Or is this a bug of molpro?
> > I need an advice: should I kill the job or let it there?
> >
> > This is the screenshot of htop.
> > https://www.dropbox.com/s/yvh4s8ttsbu7hvf/Screenshot.png
> >
> > Best regards,
> > Quan
> >
> >
> >
> >
> >
> > _______________________________________________
> > Molpro-user mailing list
> > Molpro-user at molpro.net
> > http://www.molpro.net/mailman/listinfo/molpro-user
>
>
>
> --
> Jeff Hammond
> jeff.science at gmail.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20140120/98e00cd5/attachment.html>


More information about the Molpro-user mailing list