[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gomp-discuss] "#pragma omp" and GCC
From: |
Steven Bosscher |
Subject: |
Re: [Gomp-discuss] "#pragma omp" and GCC |
Date: |
05 Feb 2003 19:22:02 +0100 |
Op wo 05-02-2003, om 18:21 schreef Pop Sébastian:
> On Wed, Feb 05, 2003 at 04:22:23PM +0100, Steven Bosscher wrote:
> > Op wo 05-02-2003, om 15:39 schreef Pop Sébastian:
> > > Steven,
> > >
> > > is it possible that you take the Open64 code
> > > and have a look at how they adapt *GCC* (yes it's
> > > the same compiler we talk about...) to parse
> > > correctly OMP pragmas before going too deep in
> > > your implementation? They implemented OMP parsing
> > > somewhere, and they did it well! Don't forget that
> > > SGI had one of the best optimizing compilers...
> >
> > Hmm, are you sure they've implemented OpenMP for C/C++? I really liked
> > the idea but "grep -irw omp *" doesn't show up anything that would
> > suggest that "their" C/C++ front ends understand "#pragma omp"...
> >
> Really strange...
> I did the same, and indeed there's nothing for parsing C/C++ OMP.
> Maybe they changed the C/C++ front-ends just before releasing their
> compiler into GPL? That is just a supposition, I have to look
> closer at the diffs below...
>
> I found in version.c the right version of their checkout:
> -const char *const version_string = "2.96 20000530 (experimental)";
> +const char *const version_string = "sgicc-1.0 {2.96 20000530
> (experimental)}";
>
> I checked some versions and just piped the result of a diff into wc
> and minimized the size: you get a minima at
> - for C:
> cvs up -d -D "2000-05-30 10:00"
> - for C++:
> cvs up -d -D "2000-06-05 23:00"
>
> I diffed the gnu directory for C and C++ front-ends.
> .diff files could be downloaded from:
> http://icps.u-strasbg.fr/~pop/sgi_gccfe.diff
> http://icps.u-strasbg.fr/~pop/sgi_g++fe.diff
>
Seems from the date of this "sgicc" that they killed the Pro64 project
just when they got started on the C/C++ front ends. It looks like the
infrastructure for OpenMP in WHIRL and their F90 compiler uses it but
they never got around to implementing it in the C/C++ front ends. What
a shame :-/
Greetz
Steven
P.S. Iis "2.96 20000530 (experimental)" the same version of "GCC" that
used to ship with Redhat?
- [Gomp-discuss] "#pragma omp" and GCC, (continued)
- [Gomp-discuss] "#pragma omp" and GCC, Steven Bosscher, 2003/02/04
- Re: [Gomp-discuss] "#pragma omp" and GCC, Biagio Lucini, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Steven Bosscher, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Lars Segerlund, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Pop Sébastian, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Steven Bosscher, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Biagio Lucini, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Pop Sébastian, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Steven Bosscher, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC, Pop Sébastian, 2003/02/05
- Re: [Gomp-discuss] "#pragma omp" and GCC,
Steven Bosscher <=
- Re: [Gomp-discuss] "#pragma omp" and GCC, Diego Novillo, 2003/02/05
Re: [Gomp-discuss] Open64's implementation of OpenMP, Steven Bosscher, 2003/02/04