Re: Regions in Program Flow

"shrey" <shreyas76@gmail.com>
23 Oct 2005 00:33:01 -0400

          From comp.compilers

Related articles
Regions in Program Flow shyamky@yahoo.com (K.Shyam) (2005-10-20)
Re: Regions in Program Flow shreyas76@gmail.com (shrey) (2005-10-23)
Re: Regions in Program Flow drizzle76@gmail.com (dz) (2005-10-26)
| List of all articles for this month |

From: "shrey" <shreyas76@gmail.com>
Newsgroups: comp.compilers
Date: 23 Oct 2005 00:33:01 -0400
Organization: http://groups.google.com
References: 05-10-135
Keywords: analysis
Posted-Date: 23 Oct 2005 00:33:01 EDT

Given a CFG this should not be hard to implement inside a compiler. If
you are looking to avoid the effort...
                  I dont know of any commercial compilers dumping anything like
this. I have read abt ORC (link is below) using such information. The
other option might be using compilers dump of call graph and loop tree
for each function. Any compiler with even some basic loop optimizations
unrolling should dump such information example ORCE,SUIF, Gcc assuming
you are looking for free solutions. Your other chance might be to find
researchers who have used such region graphs for their research. A
couple of papers that i myself have come across are


                    1. ipf-orc.sourceforge.net/region-final-version.pdf


                    2. www.ece.umd.edu/~barua/udayakumaran-CASES-2003.pdf


                    3. Some speculative precomputation paper ( there arent many
even if u search every1, but this stuff is old,)




The first one comes from the ORC guys, that I wud say is u;r best bet


Shrey


K.Shyam wrote:
> I am intrested in finding out regions of a program, those which are
> atomic units of execution. I was thinking of using the concept of
> "phases". Loop nest is also another way. Are there some other ways of
> obtaining these regions ?


Post a followup to this message

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