AMBER Archive (2006)

Subject: Re: AMBER: seg fault during test.serial

From: Lachele Foley (Lists) (lf.list_at_gmail.com)
Date: Mon Sep 11 2006 - 09:50:57 CDT


> > ==============================================================
> > cd qmmm2/crambin; ./Run.crambin_divcon
> > getRegFromUnwindContext: Can't get Gr0 from UnwindContext, using 0
> > forrtl: severe (174): SIGSEGV, segmentation fault occurred
>
> First: are you planning to use the divcon capability? If not, you can
> probably wait as this problem is looked into. We are currently evaluating
> some proposed fixes that look like optimization problems with some versions of
> ifort (see the next item).

I'm not planning to use it. If anyone else is, we have a 32-bit
machine they can use, so it's not a problem. But, I can't get the
parallel version to compile at all. I'll make that a separate post
because they don't look, to me, to be related.

> Second: what does "ifort -V" say? Sander is a fortran program, so the C/C++
> flags are not relevant here.

$ ifort -V
Intel(R) Fortran Itanium(R) Compiler for Itanium(R)-based applications
Version 8.1 Build 20040802 Package ID: l_fc_p_8.1.018
Copyright (C) 1985-2004 Intel Corporation. All rights reserved.
FOR NON-COMMERCIAL USE ONLY

> Third: if you need divcon running on the ia64, you could try lowering the
> optimization level by chaning the FOPTFLAGS variable in
> amber9/src/dcqtp/mod/Makefile, and recompiling.

I'm happy to wait, but I can try that if you're curious.

> > The test just hangs there. I have to kill the process to get out of it.
> This seems weird, since a segfault should end the job. Do you see this
> behaviour if you run test by hand, or only when it is invoked by the Makefile?

Also by hand. It won't let me cntrl-C out of it, either. It just
says "forrtl: error (69): process interrupted (SIGINT)" every time I
try, and then it continues to hang.

-- 
:-) Lachele
Lachele Foley
CCRC/UGA
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu