[Soot-list] FlowAnalysis on large graphs

[Cheng Zhang]

Hi,

Based on my experience, a lot of people may have run Soot (in its whole
program mode) on the DaCapo benchmark:
http://dacapobench.org/

I guess some of the programs are large, e.g. eclipse and tomcat.
I used to run a simple analysis (written by myself) on a quad-core machine
with 2.33GHz CPU and 16G main memory.
The analysis can be finished within hours and I think my algorithms can be
optimized to reduce the run time.

I also experienced long-running analysis run and usually had to kill the
process before it could finish.
But most of the time, I found it was due to the high time complexity of my
own algorithms.

Best,
Cheng

On Thu, Mar 29, 2012 at 1:04 PM, Karthik <reachsrk at gmail.com> wrote:

> Hi,
>
> We have defined a few ForwardFlow and BackWardFlow analyses, and are
> intending to run them on large programs.
> Enabling whole program mode and running them implies huge memory
> consumption and time. Even on small programs we have at times run into out
> of memory errors.
>
> I am interested to know the experiences of people on this list with
> running SOOT on large programs. What large programs has SOOT been used on?
> Graphs with millions of edges? what has been the experience and hardware
> requirements? What are the limits?
>
> Thanks,
>
> Karthik Ramachandra
> IIT Bombay.
> http://www.cse.iitb.ac.in/~karthiksr
>
>
> _______________________________________________
> Soot-list mailing list
> Soot-list at sable.mcgill.ca
> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.mcgill.ca/pipermail/soot-list/attachments/20120329/e08e8158/attachment.html