forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

Re: [abinit-forum] gfortran-compiled ABINIT yields two failed tests. Is this a problem?

From: "Anglade Pierre-Matthieu" <anglade@gmail.com>
To: forum@abinit.org
Subject: Re: [abinit-forum] gfortran-compiled ABINIT yields two failed tests. Is this a problem?
Date: Tue, 12 Feb 2008 09:14:41 +0100
Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:to:subject:in-reply-to:mime-version:content-type:references; b=HZqvwVHmRQfouTmbddjV160OqAnkCvolDGXl4flbS1mVSBHzfC00BJQ2ayDzXcAit6YBp3LEb92CGW7ivRuCReftLlmO56WgQXWKikdOJUGWC1rUzX7Io6/xk/RgAultRkgQ3gz3DTha/7X89HFZP8i0uWgVANvIhufmmJrS7SI=

Hi,

This is quite significant.
Yet it matters only if you intend to use PAW (and or PAW +U).
Here with gfortran 4.2.2 everything works fine on a intel core2 - FC 8/EMT64. I guess this may be due either to:
1) over optimization (have you tryed something special ?) or to instabilities of old version of gfortran.
2) also many version of gfortran coming with gcc 4.2 and 4.3 were not quite "perfect". And since we don't know exactly the date of the release (they are release aren' they ?) you use it may be the root of your problem.
The not so mature version of gfortran were all conflicting with Abinit PAW so my best guess is that you need a newer stable version of gcc. But its not quite sure.

regards

PMA

On Feb 11, 2008 6:08 PM, J. J. Ramsey <jjr19@uakron.edu> wrote:

I compiled ABINIT with gfortran 4.2.1 on Linux and gfortran 4.3 on OS X and got the same results after running "make tests_acc". (On Linux, I actually ran "make tests_acc4".) Only two tests failed, but both of them yielded NaNs. I'm not sure how significant this is, and I'd like to find out. Are we looking at a compiler bug in gfortran here?

The relevant parts of the diffs are below.

From diff.t08 in the v4/ directory:

523,524c523,524
< rms dE/dt= NaN; max dE/dt= 0.0000E+00; dE/dt below (all hartree)
< 1 NaN NaN NaN
---
> rms dE/dt= 0.0000E+00; max dE/dt= 0.0000E+00; dE/dt below (all hartree)
> 1 0.000000000000 0.000000000000 0.000000000000
530,531c530,531
< 1 NaN NaN NaN
< frms,max,avg= NaN 0.0000000E+00 NaN NaN NaN h/b
---
> 1 0.00000000000000 0.00000000000000 0.00000000000000
> frms,max,avg= 0.0000000E+00 0.0000000E+00 0.000E+00 0.000E+00 0.000E+00 h/b
534,535c534,535
< 1 NaN NaN NaN
< frms,max,avg= NaN 0.0000000E+00 NaN NaN NaN e/A
---
> 1 0.00000000000000 0.00000000000000 0.00000000000000
> frms,max,avg= 0.0000000E+00 0.0000000E+00 0.000E+00 0.000E+00 0.000E+00 e/A

997,998c997,998
< rms dE/dt= NaN; max dE/dt= 0.0000E+00; dE/dt below (all hartree)
< 1 NaN NaN NaN
---
> rms dE/dt= 0.0000E+00; max dE/dt= 0.0000E+00; dE/dt below (all hartree)
> 1 0.000000000000 0.000000000000 0.000000000000
1004,1005c1004,1005
< 1 NaN NaN NaN
< frms,max,avg= NaN 0.0000000E+00 NaN NaN NaN h/b
---
> 1 0.00000000000000 0.00000000000000 0.00000000000000
> frms,max,avg= 0.0000000E+00 0.0000000E+00 0.000E+00 0.000E+00 0.000E+00 h/b
1008,1009c1008,1009
< 1 NaN NaN NaN
< frms,max,avg= NaN 0.0000000E+00 NaN NaN NaN e/A
---
> 1 0.00000000000000 0.00000000000000 0.00000000000000
> frms,max,avg= 0.0000000E+00 0.0000000E+00 0.000E+00 0.000E+00 0.000E+00 e/A

1245,1246c1245,1246
< fcart1 NaN NaN NaN
< fcart2 NaN NaN NaN
---
> fcart1 0.0000000000E+00 0.0000000000E+00 0.0000000000E+00
> fcart2 0.0000000000E+00 0.0000000000E+00 0.0000000000E+00

From diff.t11 in the v5/ directory:

536,538c536,538
< rms dE/dt= NaN; max dE/dt= 0.0000E+00; dE/dt below (all hartree)
< 1 NaN NaN NaN
< 2 NaN NaN NaN
---
> rms dE/dt= 0.0000E+00; max dE/dt= 0.0000E+00; dE/dt below (all hartree)
> 1 0.000000000000 0.000000000000 0.000000000000
> 2 0.000000000000 0.000000000000 0.000000000000
549,553c549,553
< 1 NaN NaN NaN
< 2 NaN NaN NaN
< 3 NaN NaN NaN
< 4 NaN NaN NaN
< frms,max,avg= NaN 0.0000000E+00 NaN NaN NaN h/b
---
> 1 0.00000000000000 0.00000000000000 0.00000000000000
> 2 0.00000000000000 0.00000000000000 0.00000000000000
> 3 0.00000000000000 0.00000000000000 0.00000000000000
> 4 0.00000000000000 0.00000000000000 0.00000000000000
> frms,max,avg= 0.0000000E+00 0.0000000E+00 0.000E+00 0.000E+00 0.000E+00 h/b
556,560c556,560
< 1 NaN NaN NaN
< 2 NaN NaN NaN
< 3 NaN NaN NaN
< 4 NaN NaN NaN
< frms,max,avg= NaN 0.0000000E+00 NaN NaN NaN e/A
---
> 1 0.00000000000000 0.00000000000000 0.00000000000000
> 2 0.00000000000000 0.00000000000000 0.00000000000000
> 3 0.00000000000000 0.00000000000000 0.00000000000000
> 4 0.00000000000000 0.00000000000000 0.00000000000000
> frms,max,avg= 0.0000000E+00 0.0000000E+00 0.000E+00 0.000E+00 0.000E+00 e/A

656,659c656,659
< fcart NaN NaN NaN
< NaN NaN NaN
< NaN NaN NaN
< NaN NaN NaN
---
> fcart 0.0000000000E+00 0.0000000000E+00 0.0000000000E+00
> 0.0000000000E+00 0.0000000000E+00 0.0000000000E+00
> 0.0000000000E+00 0.0000000000E+00 0.0000000000E+00
> 0.0000000000E+00 0.0000000000E+00 0.0000000000E+00

____________________________________________________________________________________
Looking for last minute shopping deals?
Find them fast with Yahoo! Search. http://tools.search.yahoo.com/newsearch/category.php?category=shopping

--
Pierre-Matthieu Anglade

gfortran-compiled ABINIT yields two failed tests. Is this a problem?, J. J. Ramsey, 02/11/2008
- Re: [abinit-forum] gfortran-compiled ABINIT yields two failed tests. Is this a problem?, Anglade Pierre-Matthieu, 02/12/2008