Jason Swails wrote:
>> The large memory usage is probably due to the fact that both scripts are
>> setting fillratio = 4 right now so the grid setting error will never get
>> triggered. This leads to a grid too large except for small organic
>> molecules.
>>
>
> I set fillratio to 2.0 and I did not see any appreciable change.
> istrng is still the variable that will make/break the timing described
> previously
>
I was first trying to resolve the easier problem, :), i.e. Dave's
observation that some of the test cases are too large.
MJ and Jun are still working on the cause of dramatic different speeds
between the runs with and without salt ... probably the salt grid was
somehow made incompatible with the dielectric grid after many recent
revisions. If the code is working correctly, the run with salt should be
a bit faster because the solver usually converges faster.
All the best,
Ray
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Mar 30 2010 - 14:30:05 PDT
