[AMBER] 回复: Amber12 compile with "intel" err.

From: javacfish <javacfish.yahoo.com.cn>
Date: Sun, 15 Apr 2012 03:55:21 +0800 (CST)

Hi all,

I had found the final answer, maybe the Modeller (http://salilab.org/modeller/) software conflict with amber Netcdf. When I comment the modeller path and library path, the compile err will not happen again.

I am not sure what is the exactly reason, but I had solved it.


On Sat, Apr 14, 2012 at 1:55 PM, case <case.biomaps.rutgers.edu> wrote:

> [To developers: we are using a pretty old version of netcdf (3.6.2).  It
> may
> be possible that the current version (4.2) would help.  Does anyone on the
> list have experience with more recent netcdf's, especially in the Amber
> context?]

Yes -- I added the option of specifying an external NetCDF installation on
the configure line some time ago, which should bypass the NetCDF
compilation altogether and still maintain the NetCDF functionality... _if_
the Fortran module built with that external installation is usable by the
compilers you're trying to use (configure checks this).  I've been doing
this for several months, and don't actually allow Amber to build its own
NetCDF anymore (except ~half of my RC tests).

I've used both package manager-supplied NetCDFs for the GNU compiler
(MacPorts and apt-get), and built my own for the Intel compiler.

All the best,

Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
AMBER mailing list
Received on Sat Apr 14 2012 - 13:00:04 PDT
Custom Search