[Pw_forum] invalid flag

Axel Kohlmeyer akohlmey at gmail.com
Sat Nov 21 00:47:38 CET 2015


On Fri, Nov 20, 2015 at 6:07 PM, Hugo Siles <silesalvarado at hotmail.com> wrote:
> Hello,
>
> I have installed espresso-5.2.1 on fedora 22, with some minor difficulties,
> but it has been installed, I run all the tests, and as far as I could see
> all of them passed.
> However when running one of the samples from the tutorials with the
> following command:
> pw.x <Si.sample.in> Si.sample.out
> I get the following message during execution:
>
> [hsiles at localhost exercise1]$ pw.x <Si.sample.in> Si.sample.out
> Note: The following floating-point exceptions are signalling:
> IEEE_INVALID_FLAG IEEE_DENORMAL.
>
> There is an output file
>
> Does the message means any thing to worry about ??

yes and no. the messages by itself are harmless. either of the two
signaled cases can happen in normal calculations. however, they also
indicate that you may be subject to a significant performance penalty
due to how the GNU C library and compilers manage your floating point
unit. often the difference is rather small, but in pathological cases,
the difference can be substantial. since you are running on a desktop
machine, i would not worry too much. this is usually avoided by using,
e.g. a custom math library (libm) and BLAS library and also a compiler
that bypasses a lot of the
"lets-be-as-correct-as-possible-even-if-it-doesn't-make-sense-for-most-people"
nonsense the GNU people seem to love so much.

axel.

>
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum



-- 
Dr. Axel Kohlmeyer  akohlmey at gmail.com  http://goo.gl/1wk0
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.



More information about the users mailing list