[Paraview-developers] Parallel mesh reading

[sumeet kumar]

Thank You Andy,

for a quick reply.


Thanks
Sumeet

On Tue, Aug 23, 2016 at 6:58 AM, Andy Bauer <andy.bauer at kitware.com> wrote:

> The difficulty in parallel with unstructured grids is that it's tough to
> determine which cells need which points so for unstructured grids that are
> stored in a single file, all points often time are either read in on
> process 0 and broadcast out or each process reads in all points. Neither of
> these methods scale well. The advantage of multiple files here is that in
> each file the points are just the ones that are needed by the cells in that
> file. You could do a similar thing with blocks of points and cells in a
> single file. The main point here is that you need to think about an
> algorithm that has each process read in only (or nearly only) the points it
> needs in order to read in the cells in gets assigned. Also, consider
> whether or not you want your reader to be able to read in ghost cells as
> well. It's not a requirement for parallel readers but certain filter work
> better with ghost cells.
>
> On Tue, Aug 23, 2016 at 9:51 AM, sumeet kumar <sumeet.kumar507 at gmail.com>
> wrote:
>
>> I have unstructured grids
>>
>> On Tue, Aug 23, 2016 at 6:50 AM, Andy Bauer <andy.bauer at kitware.com>
>> wrote:
>>
>>> What kind of grid is it? It's much easier to read in topologically
>>> regular grids in parallel than unstructured grids.
>>>
>>> On Tue, Aug 23, 2016 at 9:47 AM, sumeet kumar <sumeet.kumar507 at gmail.com
>>> > wrote:
>>>
>>>> Hello all,
>>>>
>>>> Let's have a very large mesh in GB's/TB's that too in one file. How can
>>>> I read that file in parallel  with any arbitrary number of processors.
>>>>
>>>> How can I divide my mesh in N independent pieces? that each processor
>>>> would read. Is there any instructions for doing such kind of think in
>>>> para-view.
>>>>
>>>> Is reading multiple files in parallel more efficient or reading one
>>>> file in parallel more efficient in para-view. In either of these how I
>>>> should design my output to be most efficient to be read by ParaView in
>>>> Parallel.
>>>>
>>>>
>>>> Sumeet
>>>>
>>>> --
>>>>
>>>> Sumeet Kumar Sinha
>>>> Graduate Student
>>>> Phone: (+1)  <%28%2B91%29%209910516219>5306018271
>>>> Website : http://www.sumeetksinha.com/
>>>>
>>>> _______________________________________________
>>>> Powered by www.kitware.com
>>>>
>>>> Visit other Kitware open-source projects at
>>>> http://www.kitware.com/opensource/opensource.html
>>>>
>>>> Search the list archives at: http://markmail.org/search/?q=
>>>> Paraview-developers
>>>>
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://public.kitware.com/mailman/listinfo/paraview-developers
>>>>
>>>>
>>>
>>
>>
>> --
>>
>> Sumeet Kumar Sinha
>> Graduate Student
>> Phone: (+1)  <%28%2B91%29%209910516219>5306018271
>> Website : http://www.sumeetksinha.com/
>>
>
>


-- 

Sumeet Kumar Sinha
Graduate Student
Phone: (+1)  <%28%2B91%29%209910516219>5306018271
Website : http://www.sumeetksinha.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview-developers/attachments/20160823/1d82f77f/attachment.html>