This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH]: Speed up dominance frontiers computation
On Mon, 2005-01-17 at 16:54 +0100, Paolo Carlini wrote:
> Paolo Carlini wrote:
> > Since you explicitly mention Harvey, isn't a short bibliographic
> > reference in order?
> Guessing *wildly* , this one
That's actually a *dominance* algorithm, not one for dominance
You'll note that algorithm, along with updated numbers, is included in
Harvey's thesis (again, http://www.cs.rice.edu/~harv/thesis.pdf), which
is from 2003.
I didn't bother to implement the iterative dominance algorithm because
the numbers he produced aren't significantly better than the ones for
our algorithm. However, you'll note the df algorithm numbers *are*
siginficantly better (30-45%).