Re: [abc] cflow call graph timings

On Fri, Apr 15, 2005 at 07:20:51PM -0400, Prof. Laurie HENDREN wrote:
> I think these are worth putting in the paper. What do you think?

I don't think so. I think they dilute the happy story too much.

Also, they open up lots of questions we haven't answered. We don't know
why sablecc is so different. We don't know whether CHA could be
implemented in BDDs more efficiently.

I think these numbers are interesting to us to begin answering these
questions, but in the paper, they would draw attention away from the
very nice, finished results we already have in there.

One thing that I think *is* worth mentioning, and that goes along with
the story, is the precision difference on sablecc and weka (I think I
sent the details in an earlier e-mail).

Ondrej

>
> Laurie
>
>
> +-------------------------------------------------------------+
> | Laurie Hendren, Professor, School of Computer Science |
> | McGill University |
> | 318 McConnell Engineering Building tel: (514) 398-7391 |
> | 3480 University Street fax: (514) 398-3883 |
> | Montreal, Quebec H3A 2A7 hendren@cs.mcgill.ca |
> | CANADA http://www.sable.mcgill.ca/~hendren |
> | http://wwww.sable.mcgill.ca http://aspectbench.org |
> +-------------------------------------------------------------+
>
> On Fri, 15 Apr 2005, Ondrej Lhotak wrote:
>
> > I've split out the cflow timings into encoding the call graph in bdds
> > and actually doing the analysis and changing advice lists. Here's the
> > breakdown. All times are in ms.
> >
> > Notice that: 1) encoding the CG is the bulk of the time 2) analysis
> > itself is quite quick, except in sablecc 3) bigger CG (CHA) takes longer
> > to encode 4) analysis takes a bit longer on bigger CG
> >
> > Analysis/advice lists:
> > PTA CHA
> > ConaSim 395 668
> > sablecc 90138 101137
> > ants 207 128
> > conastack 152 174
> > figure 73 100
> > quicksort 67 95
> > lodsim 576 1293
> > weka 1917 2153
> >
> > Encoding CG in BDDs:
> > PTA CHA
> > ConaSim 19919 47418
> > sablecc 155913 211815
> > ants 24970 61446
> > conastack 19315 57910
> > figure 17395 44657
> > quicksort 19222 45051
> > lodsim 27903 148714
> > weka 82384 207530
> >
> >
> >
> >
>
>
Received on Sat Apr 16 04:08:58 2005