Re: [AMBER] Problem with AMBER20 compilation using Intel/20.4 compiler

From: David A Case <david.case.rutgers.edu>
Date: Tue, 9 Mar 2021 12:15:56 -0500

On Mon, Mar 08, 2021, Samuel Mohan Dass, Premila wrote:
>
>ifort: error #10105: /jet/packages/intel/compilers_and_libraries_2020.4.304/linux/bin/intel64/fortcom: core dumped
>ifort: warning #10102: unknown signal(21370896)
>ifort: error #10106: Fatal error in /jet/packages/intel/compilers_and_libraries_2020.4.304/linux/bin/intel64/fortcom, terminated by unknown
>compilation aborted for /ocean/projects/mcb200041p/premila/amber20_src/AmberTools/src/sander/ips.F90 (code 1)

Two points of information here:

1. For me, with intel 2019.5 compilers and the current development codes,
the CMake build looks fine. If anyone is using 2020 Intel compilers, you
might note if you have or have not seen a compiler core dump when trying to
compile ips.F90.

2. Premila is working on XSEDE, and has been assigned to Bridges-2, which
looks like it's not quite ready for prime time. Among other things, the
only GNU compilers it makes available is gcc10, which won't work for Amber.
Bridges itself has lots of modules for earlier versions, but the Exceed
website doesn't even mention Bridges-2 yet. So: if anyone can get Amber
running on Bridges-2, we should post hints for that.

Premila: are you using the master branch from gitlab.ambermd.org, or the
released Amber20 tarballs? The former would be better, although ips.F90
has not changed since Feb. of 2016. If Intel compilers can't handle that
source file, I would have thought we would have found out before now....

...thx...dac

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Mar 09 2021 - 09:30:02 PST
Custom Search