AMBER Archive (2009)

Subject: Re: [AMBER] how to get hold of older versions of MKL ?

From: Jason Swails (jason.swails_at_gmail.com)
Date: Mon Oct 19 2009 - 05:54:45 CDT

Hi,
I believe that the three MKL libraries that you need are the following for a
64-bit machine (uname -m = x86_64)

libmkl_core.a
libmkl_intel_lp64.a
libmkl_sequential.a

If you have a 32-bit machine (uname -m = i686):

libmkl_core.a
libmkl_intel.a
libmkl_sequential.a

So in your CFLAGS/FFLAGS make sure that you are sourcing those libraries.
 libmkl_lapack.a I believe is part of libmkl_intel(_lp64).a, so you
shouldn't explicitly need that one. Also, make sure that your environment
variables are appropriately set up (i.e. you LD_LIBRARY_PATH, etc.). The
intel MKL provides some scripts to do that. Add to your .bashrc the
following couple lines:

source $MKL_HOME/tools/environment/mklvars(architecture).sh (where
architecture is 32 for 32-bit, em64t or 64 for 64-bit machines).

I also don't remember if there are certain flags unique to old versions of
MKL that the configure script puts into the CFLAGS or FFLAGS (specifically
-lvml). (If the current version of the configure scripts cannot
discriminate between MKL versions <= 9, future configure scripts will be
able to). If -lvml is listed in any of the compiler flags, you have to take
that out for MKL versions 10 or higher. I would do the same for any other
flag that it complains about not recognizing, but to my recollection there
shouldn't be any other.

Hope this helps!

Best of luck,
Jason Swails

On Mon, Oct 19, 2009 at 4:59 AM, Nahoum Anthony <nahoum.anthony_at_strath.ac.uk
> wrote:

> Dear all,
>
>
>
> I'm trying to compile Amber tools 1.2 and Amber 10 but run into problems
> with Intel's Math Kernel Library. I've just downloaded version 10.2.2.025
> and it's missing libmkl_lapack.a which the compilation of Amber Tools
> requires when MKL_HOME is set as environment variable. How can I get an
> older version of the MKL ? is there another work around this, given that I
> want to use MKL ?
>
>
>
> Thanks,
>
>
>
> Nahoum
>
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
> 

-- 
---------------------------------------
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber