VMD-L Mailing List
From: John Stone (johns_at_ks.uiuc.edu)
Date: Fri May 01 2009 - 10:36:23 CDT
- Next message: Sasha Buzko: "Re: autopfs: unknown atom type error when processing ATP"
 - Previous message: Jim Pfaendtner: "Re: VMD on Cray/CNL"
 - In reply to: Jim Pfaendtner: "Re: VMD on Cray/CNL"
 - Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
 
Jim,
  Can you send me the diffs you made to allow compilation on the
Cray compute node Linux?  I would like to make it easier for
the next person to do this type of build by adding a new VMD
configuration type based on yours. :-)
Cheers,
  John
On Fri, May 01, 2009 at 02:56:02PM +0200, Jim Pfaendtner wrote:
> Hi John,
> 
> Thanks for your prompt response.  I did everything as you said and it  
> seems to be working (based on several test jobs).   I can now call the  
> vmd _LINUXAMD64 executable (now statically linked) with the CNL job  
> submission program aprun.
> 
> I had to tweak my submission script a bit to set up all of the  
> environment variables properly.  Also, the compiled vmd program needs  
> everything in the installation directory, so the entire contents have  
> to be transferred over to scratch if one is working on a system that  
> does not allow access to the home directories from compute nodes (as  
> in my case).
> 
> One other note is that I also had to download and compile the static  
> version of netcdf which was required for the vmd plugins.  In the end  
> I built vmd using:
> 
> ./configure LINUXAMD64 TCL NETCDF STATIC
> 
> and I only needed minor adjustments to the Makefile to set paths, etc.
> 
> Jim
> 
> 
> 
> On May 1, 2009, at 8:20 AM, John Stone wrote:
> 
> >
> >Hi,
> > Yes, it is possible to build VMD with static linkage, but it
> >will be somewhat tricky.  For your application in particular, you'll
> >first have to have a statically linkable version Tcl compiled and
> >ready for use.  Once you've got Tcl compiled, then you'll need to
> >build the VMD plugins for static-linkage only.  You can probably hack
> >the 64-bit linux compile target and simply disable the shared object
> >plugin builds.  Once you get the plugins compiled as a static library,
> >building VMD itself with only static linkage will just be a matter of
> >tweaking a few options with the 'configure' script, disabling all of
> >the optional features that you don't need.  I'd start with one of the
> >"linux.text" configurations in vmd/Makefile, and then tweak from  
> >there.
> >I don't currently have convenient access to an XT for compiling VMD,
> >otherwise I'd offer to help do the build for you.  Let us know if you
> >have questions.  If anyone else has compiled other non-trivial  
> >packages
> >for the Cray XT, feel free to chime in...
> >
> >Cheers,
> > John Stone
> > vmd_at_ks.uiuc.edu
> >
> >On Fri, May 01, 2009 at 08:03:07AM +0200, Jim Pfaendtner wrote:
> >>Dear VMD-l,
> >>
> >>For a variety of reasons, I need to run VMD on a Cray XT machine with
> >>CNL from the compute nodes.  That is, I have to call VMD using the
> >>aprun command within my job submission script.  The main reason is  
> >>the
> >>enormous amount of post-processing I want to do on some very large
> >>systems is not suitable for the login nodes which do not have a large
> >>amount of memory and a 15 minute limit for user-run executables.
> >>
> >>This currently does not work when I attempt it with the  
> >>vmd_LINUXAMD64
> >>executable (this executable works fine on the front end).  At a bare
> >>minimum the executables for CNL have to be statically linked and the
> >>VMD binaries I downloaded seem to be dynamically linked.  I have no
> >>idea what else might be required.
> >>
> >>1) Is it possible to build a VMD binary that will run using aprun on
> >>CNL?
> >>
> >>2) If so, could someone point me in the right direction to get  
> >>started?
> >>
> >>Thank you,
> >>
> >>Jim
> >
> >-- 
> >NIH Resource for Macromolecular Modeling and Bioinformatics
> >Beckman Institute for Advanced Science and Technology
> >University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
> >Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349
> > WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078
-- NIH Resource for Macromolecular Modeling and Bioinformatics Beckman Institute for Advanced Science and Technology University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801 Email: johns_at_ks.uiuc.edu Phone: 217-244-3349 WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078
- Next message: Sasha Buzko: "Re: autopfs: unknown atom type error when processing ATP"
 - Previous message: Jim Pfaendtner: "Re: VMD on Cray/CNL"
 - In reply to: Jim Pfaendtner: "Re: VMD on Cray/CNL"
 - Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
 



