In defense (sort of) of our f90 compiler:
The current shipping version (1.2) is fairly solid, but (yes, this is an
admission by a "Sun insider") pretty danged slow. Admitted and confessed.
The upcoming version (2.0) is much faster, because we have bolted the regular
Sun backend onto it; it is currently almost as fast as the f77 compiler for
FORTRAN 77 code. However, that's pretty major surgery that wasn't completed too
long ago, so there are still some stitches showing. We are squashing bugs every
day on f90, and the development team has made bugs their #1 priority from now
until we ship early next year. I have excellent reason to believe that f90 will
be solid and fast when it ships.
So, the second confession is that the EFT (Beta) version of f90 2.0 is not as
solid as we'd like. Generally, we have seen it doing better than Glenn
indicates, but we have seen results like his. Please trust that we are in
maximum bug-killing mode.
I'd nonetheless like to encourage everyone on this list to download the latest
bits from http://access1.sun.com/workshop5.0ea (they were updated October 23)
and test them with your code, then let us know the results (good or bad).
Thanks,
Bill Moffitt
Glenn Carver wrote:
>
> At 6:37 pm +0000 17/11/98, Phillip Helbig wrote:
> >My impression would be that the solaris compiler is not worth messing
> >with. This has been my experience and SUN people have admitted publicly
> >that it is not really to be taken seriously. So what's the point in
> >bending over backward to make your stuff work with it?
> >
> >Supposedly, the next version (now downloadable in beta test, I believe)
> >is much better. Maybe you should try it with that?
>
> Well, it's true it fixed some problems we'd been having but the compiler
> crashed on 3 separate errors on relatively simple code within my first 10
> mins of testing -- maybe I was unlucky, but interestingly it was the most
> f77 like of the code I had. The web reference is
> http://access1.sun.com/workshop5.0ea/.
>
> ----
> Dr. Glenn Carver, Senior Research Associate,
> Centre for Atmospheric Science, Chemistry Dept., Cambridge University, UK
> [log in to unmask] http://www.atm.ch.cam.ac.uk/~glenn/
>
> "Genius is one percent inspiration and ninety nine
> percent perspiration" Thomas Edison.
--
======================================================
Bill Moffitt Product Manager, Performance WorkShop
[log in to unmask] (650) 786-8886
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|