[J3] Performance Portability and Fortran: Making Fortran cool again

[Keith Bierman]

On Thu, Jan 17, 2019 at 11:55 AM Ondřej Čertík via J3 <
j3 at mailman.j3-fortran.org> wrote:

>
>
> On Thu, Jan 17, 2019, at 11:39 AM, Gary Klimowicz wrote:
>
> >
> > And if I understand Ondrej's comment correctly, I agree that the
> > optimization hints belong at configure time or runtime, and not be part
> > of source code itself.
>
> Yes.
>
>
Well, this seems easy to say ... but not so clear (at least in my
experience) in practice.

A very common (indeed ancient) thing to assert is that specific loop(s) are
(despite language permissiveness) really iteration independent ("IVDEP").
In fact, such assertions are (again, in my experience) often not generally
true (may depend on various runtime parameters) ... other things I may wish
to let the compiler know may include things like "the loop must be a
multiple of 3" ... such things make a lot more sense in the context of the
source than in some external configuration file.

That we keep "running in circles" with adding directives (annotations,
whathaveyou), compiler options and sometimes rich configuration files ...
in addition to language constructs isn't a reflection of an inability to
learn ... but that commercially significant usecases exist for all of these
modes of operation ... and commercially successful vendors keep finding
this out generation after generation ;>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20190117/523fab79/attachment.html>