[FORTRAN PATCH] PR 30404: Rewrite nested FORALL mask expansion
Steve Kargl
sgk@troutmask.apl.washington.edu
Tue Jan 16 19:08:00 GMT 2007
On Tue, Jan 16, 2007 at 11:42:56AM -0700, Roger Sayle wrote:
>
> On Mon, January 15, 2007 5:02 pm, Steve Kargl wrote:
> > On Sat, Jan 13, 2007 at 04:04:05PM -0700, Roger Sayle wrote:
> >> If someone could do the usual benchmarking and conformance testing,
> >> I'd be interested to know if there's a positive or negative impact.
> >
> > Roger, I scanned my work/ directory where I store the Fortran
> > code I've snarfed from the internet. With over million lines of
> > code, there is exact ZERO ocurrences of FORALL. :(
>
> Doh! I wonder how much of this is a chicken-and-egg problem? If most
> fortran compilers currently generate inefficient code for these constructs
> and are unable to take advantage of the vectorization benefits, I suspect
> that many programmers would use alternative idioms.
>
I think you're right about the chicken-and-egg problem. To use
FORALL in anything but its simplest form requires a fairly
good analysis on the dependencies in the code. I suspect most
Fortran programmer do not spend the necessary timei to learn FORALL.
They either continue to use do-loops or they turn to OpenMP.
I know you plan to look into some optimizations with FORALL,
so I'll continue to try to locate real world apps for testing.
--
Steve
More information about the Gcc-patches
mailing list