On Tue, Jul 24, 2007, Carlos Simmerling wrote:
>
>
> I think we can keep the LES-style NEB/PIMD completely separate from the
> multisander one. you're right that LES is overly complicated, so many ifdefs
> and so on. we could think of converting the LES code into more of a partial
> multisander, and then have specific options depending on whether it is
> LES/NEB/PIMD/etc but that would take a little work, I don't think I could do
> much for amber10.
So: is anyone (Kim, Ross, DaveM?) using "LES-style NEB/PIMD" right now? My
tentative feeling is that we should concentrate on getting the multisander
stuff going (having lots of experience, good documentation, clean code, remove
unnecessary communications, etc.), and put of any LES-based algorithms for
Amber 11.
> we were thinking last week of extra communicators. Since this
> will be very useful in the future and is such a pain to do....
Has anyone tried to more or less replicate the code starting at line 196
of multisander.f ("Create a communicator for each group of -ng NumGroup
processors:"), to make a second set of communicators (say for EVB)? I wonder
if we are not assuming that this is harder to do than it really is. To be
clear: I've not tried this myself: it would take a little time, but once
everything is set up at the top (multisander) level, things below that could
be pretty simple(?).
...dac
Received on Wed Jul 25 2007 - 06:07:35 PDT
