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: SIMPLE: A language-independent tree IR


On Jan  6, 2002, Richard Henderson <rth@redhat.com> wrote:

> On Sun, Jan 06, 2002 at 11:48:11PM -0200, Alexandre Oliva wrote:
>> If SIMPLE is what we're aiming at, let's figure out how to
>> represent it in RTL.  I don't see the point of creating yet another
>> set of data structures, management functions, etc.  It's a waste of
>> work.

> Err, we're not.  It's trees.

It makes a lot more sense now.

However, the similarity between SIMPLE and RTL makes me wonder how
much of gain we're really going to have by introducing such
optimizations on trees, instead of just converting to a MI RTL and
applying optimizations there.

The advantage would be that we might be able to take advantage of
existing optimization passes, that already work in RTL, and introduce
new optimization passes that are intermingled with the existing ones,
instead of having to re-implement many of the existing optimizations
such that they work on trees.

I mean, if we're flattening out trees as it appears to me that SIMPLE
does, it seems to me that it's just RTL written in a different way.
So why not bite the bullet and go with MI RTL, even if slightly
augmented so as to contain data currently unavailable in the RTL
level.

Or am I totally missing the point, as usual? :-D

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me


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