Re: [AMBER-Developers] TOT PMEMD cuda_serial unbuildable?

From: Scott Le Grand <varelse2005.gmail.com>
Date: Tue, 8 Mar 2022 13:02:25 -0800

So I give up, but configure_cmake.py -cuda worked...

But now, where do I edit out sm revisions for which I do not wish to build?
It's easy with configure, just hit config.h afterwards. I need to get the
compile link debug cycle down to something manageable. My work tree is 28
seconds. I'll be happy with 2 minutes or less here.



On Tue, Mar 8, 2022 at 12:51 PM David A Case <david.case.rutgers.edu> wrote:

> On Tue, Mar 08, 2022, Scott Le Grand wrote:
>
> > I still cannot build pmemd.cuda_SPFP
>
> The frustrating thing is that I and others do this many times a week, and
> the
> CI process does this several times a day (at present). What is hard to
> figure
> out is how your setup is different from a more standard one.
>
> I really recommend (a) using the clean_build script to make sure that you
> have
> a pristine build environment; (b) bite the bullet and just do a complete
> install -- you only have to do that once -- after that, fiddling with pmemd
> will build very quickly. By all means, type "./configure_cmake.py --help"
> and go from there if it is more to your liking.
>
> An historic problem is that pmemd and AmberTools are very closely
> intertwined, so it can be tricky to just say "make pmemd.cuda_SPFP" from
> bare metal. But once you've installed once (which gets things like netcdf
> into good shape), you should just be able to build a single program; I do
> that
> a lot. It's always great when thing work, but less convenient when
> something
> breaks.
>
> ....dac
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Mar 08 2022 - 13:30:03 PST
Custom Search