Re: [AMBER] Confirm compiling Amber10/Tools1.3 breaks with intel v 11.1?

From: case <>
Date: Wed, 17 Mar 2010 12:01:13 -0400

On Tue, Mar 16, 2010, Trevor Gokey wrote:
> Hey everyone, I'm getting no luck compiling with the newest intel
> compilers. Everything but pmemd compiles but all of the tests for
> the othe r programs fail somewhere. Pmemd complains about a global
> variable not havi ng an explicit type in pme_setup.fpp. I'm just
> assuming to avoid v1= 1.1

I can't confirm this. AmberTools 1.3 shows no problems for me with icc/ifort
11.1 (069). Sander looks fine if MKL is turned off; some hand-editing of
config.h is required if MKL is desired. The pmemd problem is one we know
about--my guess is that a simple, two-line patch will fix things:

--- pme_setup.fpp 10 Dec 2009 02:39:07 -0000 10.1
+++ pme_setup.fpp 11 Mar 2010 16:28:10 -0000 10.2
.@ -403,8 +403,8 @@
 subroutine vdw_correct_setup
- use pme_force_mod
- use prmtop_dat_mod
+ use pme_force_mod, only : gbl_nvdwcls
+ use prmtop_dat_mod, only : natom, atm_iac
   implicit none
Of course, run the test cases to make sure things are OK. For most
simulations, you won't loose much by skipping MKL. If this seems OK, we
should post and Amber10 patch.

Of course, you can also back out the compiler, and/or wait for Amber11, where
we are concentrating our testing efforts for the moment.

...hope this helps....dac

AMBER mailing list
Received on Wed Mar 17 2010 - 09:30:02 PDT
Custom Search