This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: GCC vs. Open64


2007/6/25, Basile STARYNKEVITCH <basile@starynkevitch.net>:
Jiahua He wrote:
> Hi All,
>
> I am looking for an open source compiler, such as GCC, Open64, Scale
> and SUIF, as the infrastructure of my research. Though I have read
> some of their documents, they are still complicate to me and it is
> difficult to decide which one is better. Can you please help me to
> compare these compilers in terms of the following requirements?

Very few people could answer you precisely and with expertise,
because understanding one compiler is already a big work
(so deeply knowing several compilers is not very reasonable for a single
person).

You probably want to work on some internal representation.
In several compilers (in particular in GCC with its GIMPLE or GIMPLE/SSA)
the internal representation is (almost) not tied to a specific source language.
In other words, your request might not be very Fortran specific.

You also might consider adding LLVM to your list of candidates
(even if I don't know if there is a working gcc based frontend to LLVM for Fortran).

AFAIU several compilers fits the bill quite nicely
(since they have a source-neutral internal representation), including GCC.

Maybe a relevant indicator is the size of the developer community.
In that case GCC is probably one of the best.

Good luck.


-- Basile STARYNKEVITCH http://starynkevitch.net/Basile/ email: basile<at>starynkevitch<dot>net mobile: +33 6 8501 2359 8, rue de la Faïencerie, 92340 Bourg La Reine, France *** opinions {are only mines, sont seulement les miennes} ***


Thanks for quick reply!


I understand IR is language-independent. But I do need a Fortran front
end to work on my Fortran codes. That is why I didn't include LLVM in
my list. As for as I know, LLVM doesn't have a front end for Fortran
yet.

The size of developer community is surely one of the indicators. But I
am more interested in another one: how many research projects are
based on GCC, especially on program analysis instead of optimization?
This might be an easier but still helpful question. :-)

Jiahua


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]