[m-dev.] Implicit Parallelism test cases.

[Peter Wang]

On 2008-10-16, Paul Bone <pbone at csse.unimelb.edu.au> wrote:
> On Wed, Oct 15, 2008 at 11:03:02PM +1100, Paul Bone wrote:
> > 
> > I'm looking for test cases for implicit parallelism, namely programs
> > roughly the size of icfp2000 or smaller, that have some implicit
> > AND-parallelism and don't currently contain any parallel conjunctions.
> > 

Hi Paul,

~wangp/mershared/final_benchmarks/icfp2000 seems to be my modified copy.
It accumulates pixel values per row then writes out a row at a time.
You could modify it to accumulate rows of rows so there is no I/O in the
main loop.

> I recall that Tom Conway used a monte carlo simulator to test parallel
> conjunctions in his Phd thesis.  Do we have a (working) copy of this
> program?

One of these:

~wangp/mershared/benchmarks/monte
~wangp/par_monte

> If I cannot find any I might write something like a Mantlebrot image
> generator.

extras/graphics/mercury_allegro/samples/mandel might be helpful.

I don't know if it would help, but liveness.m and proc_gen.m were once
modified to work in parallel (see detect_liveness_preds_parallel and
generate_code_parallel).  The & were changed to , for bootstrapping
reasons but were never put back.

Peter

--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at csse.unimelb.edu.au
Administrative Queries: owner-mercury-developers at csse.unimelb.edu.au
Subscriptions:          mercury-developers-request at csse.unimelb.edu.au
--------------------------------------------------------------------------