(j3.2006) parameterized derived types

[Bill Long]

On Dec 11, 2015, at 7:52 AM, Tom Clune <Thomas.L.Clune at NASA.GOV> wrote:

> 
>> On Dec 11, 2015, at 12:53 AM, Keith Bierman <khbkhb at gmail.com> wrote:
>> 
>>> , as a user, I would have little sympathy for another user that fell into this abyss. 
>> 
>> If the application runs on a mix of processors, say GPUs and x64 and power processors won?t it be fairly easy to fall into the trap (we set by allowing fp  in these contexts)?
> 
> I await a plausible example.  Certainly it is easy to create an artificial example, but using FP operations to determine array lengths is vanishingly rare in my experience.  

Likewise.  Further, the problem exists only when the computed argument to INT ( ) is close enough to an integral value that rounding would tip the integer result the wrong way.   This is a theoretical problem in the standard, but in practice it is in the epsilon**2 probability range.   I have never seen a customer problem report where this was the cause.

Cheers,
Bill


>  And I don?t dispute that exotic HW would make it easier to demonstrate the problem.   And if the expression is vulnerable to roundoff, it begs the question as to what the intent really was.
> 
> Note.  I?m not denying the possibility that someone will show me a plausible example.  And, I?m fully ready to reverse my position as a consequence.   I suppose mostly, I?m challenging others to come up with that plausible example.
> 
> Cheers,
> 
> - Tom
> 
> 
>> 
>> khbkhb at gmail.com | kbiermank (AIM) | khbkhb123 at icloud.com (iMessage)
>> 
>> Sent from my iPad
>> 
>> 
>> _______________________________________________
>> J3 mailing list
>> J3 at mailman.j3-fortran.org
>> http://mailman.j3-fortran.org/mailman/listinfo/j3
> 
> Thomas Clune, Ph. D. 	<Thomas.L.Clune at nasa.gov>
> Software Infrastructure Team Lead
> Global Modeling and Assimilation Office, Code 610.1
> NASA GSFC		
> MS 610.1 B33-C128
> Greenbelt, MD 20771
> 301-286-4635
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> _______________________________________________
> J3 mailing list
> J3 at mailman.j3-fortran.org
> http://mailman.j3-fortran.org/mailman/listinfo/j3

Bill Long                                                                       longb at cray.com
Fortran Technical Support  &                                  voice:  651-605-9024
Bioinformatics Software Development                     fax:  651-605-9142
Cray Inc./ Cray Plaza, Suite 210/ 380 Jackson St./ St. Paul, MN 55101