Interest article

[Steve Kargl]

On Tue, Dec 14, 2010 at 09:24:49AM +0100, Paul Richard Thomas wrote:
> Dear Steve,
> 
> > "gfortran developers have made significant strides in efficient
> > ?implementation, approaching the efficiency of C or f77 in more
> > ?cases than ifort."
> 
> Considering that this is only version 4.3 that they are talking about,
> it is good news.  It would be interesting to know how much of the
> credit that the gfortran developers can take and how much is due to
> gcc developments.

I suspect we can take only a small portion of the credit.  The article's
main topic is the vectorization capability of various Fortran compilers.

> >
> > But, there appear to be some warts: "All good except gfortran."
> 
> What does this mean?  The benchmark failed, gfortran was slowest, or.....?

In the few places that this phrase appears, I intrepret this
to mean that gfortran had problems with vectorizing the code.
It is not clear whether the phrase also means that gfortran
produced wrong results or was horrendously slow.

> > PS: Most comments apply to gfortran 4.3, so gfortran may have
> > addressed some of the problems mentioned.
> 
> Indeed, the article says as much.
> 

Well, the good news is that I have 4.3, 4.4, 4.5, and 4.6 
installed on troutmask.  At the end of the article, it appears
that the source code for the benchmark is available.  Perhaps,
in my infinite free time, I'll see how gfortran does on the
newer versions.

-- 
Steve