Re: amber-developers: Re: amber10 test failures with g95/x86_64

From: Kim F. Wong <kfwong.hec.utah.edu>
Date: Wed, 27 Feb 2008 16:10:24 -0700

Freaking awesome ... it works right out of the box in my case. Excellent.

-Kim

David A. Case wrote:
> On Wed, Feb 27, 2008, Kim F. Wong wrote:
>
>
>> Would someone help me get gfortran & mpich 1.2.7 working on my laptop:
>>
>
> I can't help with the exact command above, but I have finally done what
> was threatened long ago: put a version of lam into the Amber10 tree.
>
> So, if you "make clean" and then "./configure -lam <compiler>" you should
> get instructions on what to do next (i.e. run configure_netcdf and
> configure_lam). This followed by a "make parallel".
>
> In my experience, lam is the most portable of the mpi's, and is certainly good
> enough to "just get something working on my laptop".
>
> Notes:
>
> a. For right now, before the above, go to $AMBERHOME and type "ln -s exe
> bin". (I'll make this automatic soon).
>
> b. Of course, even this may not work on Kim's laptop. But, finding out why
> and fixing the scripts will then pay dividends for everyone, because we will
> have a hardwired, reproducible and testable installation, instead of having
> everyone install a completely different MPI, then report problems to the
> mailing list. Our installation will guarantee that the same compiler is
> used, that all libraries and execuatbles are local to the amber10 tree, etc.
>
> c. Anyone wishing to help can explore configure options for lam. Look at the
> "configure_lam" script that gets created, and see if there are other packages
> we could leave out -- the idea is to only build the stuff we really need,
> hence avoiding unnecessary dependencies. Also, the current script won't build
> a static MPI version...I hope someone can figure out why and fix it.
>
> d. Of course, it could be that LAM was/is the wrong choice, but let's give it
> a try for a while. One of my points is that lam (alone among things I have
> tried) will compile and run on cygwin, which is one (fairly severe) test of
> portability.
>
> ...comments and help welcome here....thanks...dave
>
>
>
Received on Sun Mar 02 2008 - 06:07:15 PST
Custom Search