This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Flow analysis and optimization with computed gotos
- To: Richard Henderson <rth at redhat dot com>
- Subject: Re: Flow analysis and optimization with computed gotos
- From: Michael Matz <matzmich at cs dot tu-berlin dot de>
- Date: Wed, 6 Dec 2000 02:48:39 +0100 (MET)
- cc: Joern Rennecke <amylaar at redhat dot com>, Brad Lucier <lucier at math dot purdue dot edu>, gcc at gcc dot gnu dot org, feeley at iro dot umontreal dot ca, hosking at cs dot purdue dot edu, Andrew Macleod <amacleod at redhat dot com>
Hi,
On Tue, 5 Dec 2000, Richard Henderson wrote:
> > We can still pretend to work on the entire graph, and thus get all
> > subgraphs.
>
> But part of the problem with Brad's test cases is the relatively
> complete connectivity of the graph. For such examples it would
> seem to be most efficient to actually operate on the subgraphs.
Brad has such cases, yes, but this particular one only has two computed
goto's in the whole function, so there it's not thaat bad. Not, that it
wouldn't be good to be able to process subgraphs ;-) But for being able to
really process "subgraphs" (for whatever that means in this context), and
still have connections between the subgraphs (Or do you mean really
isolated graphs?) I think, proper structural analysis will be needed. And
until then it really seems to be enough to simply ignore abnormal critical
edges. Or follow Jeff's idea.
Ciao,
Michael.