sieve-dev globalOrder

[Matthew Knepley]

On Tue, Oct 7, 2008 at 10:58 PM, Shi Jin <jinzishuai at gmail.com> wrote:
> Hi there,
>
> I did a final memory scan on my code and it turned out that everything is in
> good shape with the exception of the globalOrders.  For example, for a code
> using 240,000 free nodes, the total code memory cost is about 1.2GB, in
> which the globalOrders take about 500MB. I understand that something can be
> done to the globalOrders  too, just as the sieve to isieve transition, which
> will significantly reduce its memory cost. This is good but I also
> understand it is probably not something to have in the near future. So I am
> thinking about a workaround. In my case, my 3 components of the velocity
> vector share the same structure (at least for the time being, BC is for all
> three components). So the globalOrder on this three sections would be
> identical. I am just wondering if there is a way to let them share the
> memory usage for globalOrders. To do that, we need something like
> setGlobalOrder() or even setFactory(). I guess it would be fairly
> straightforward to do. Could you please let me know if you can add it for
> me. Thank you very much.

Sure I can add this, I just need to understand a little better. The
global orders
are divided by mesh and name, and I assume you would like different sections
to share the same global order. You can just call getGlobalOrder() with the same
name and you will get back the same order. I can add a setGlobalOrder() so that
names can be aliased.

   Matt

> --
> Sincerely,
> Shi Jin, Ph.D.
> http://www.ualberta.ca/~sjin1/
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which
their experiments lead.
-- Norbert Wiener