Re: PRE information needed

"Vikram.V.Asrani" <vikram@eecs.umich.edu> wrote:


> I am planning to do my compiler project on Partial Redundancy
> Elimination. I have only a brief idea on what this topic covers and
> am trying to get more information on this topic. I believe that there
> are a couple of algorithms, one of which is a standard one but is more
> restrictive. I am searching information about the other algorithm
> which I believe is called PRE.


The initial paper is by Morel & Renvoise. The later revision is called
Lazy Code Motion by Knoop, Ruthing, & Steffen. The "standard"
algorithm these days is the improvement to the KRS approach by
Dreschler & Stadel.


The D&S paper can be found in SIGPLAN notices 28:5 May 1993.


There's also a good summary of this approach in Morgan's text, and an
interesting variation that combines PRE with Global Value Numbering in
L. Taylor Simpson's PhD thesis, titled "Value-Driven Redundancy
Elimination"


Also, Preston Briggs did a nice summary of PRE. You can find the
Briggs and Simpson stuff if you poke around at www.cs.rice.edu looking
for the Massively Scalar Compiler Project. There's a page where all of
the papers are available in postscript form.


--
John McEnerney (johnmce@texas.net)

Return to the comp.compilers page.
Search the comp.compilers archives again.