[ncl-talk] Parallel NCL

[Mary Haley]

Zhifeng,

The parallel NCL project that you're thinking of is likely ParVis, which
was a specialized project to process certain CESM files, with only a subset
of analysis routines parallelized.  I don't think this will be appropriate
for your requirements.

One thing we are doing slowly is introducing some OpenMP parallelism on
internal algorithms (graphical and computational), and we hope to elevate
this activity after we hire two new developers in the NCL team.

We are also investigating ways to use Python tools to introduce
parallelism, using the mpi4py module. The Application Scalability and
Performance Group at NCAR has written two Python-based parallel tools:

   - PyReshaper - takes time slice data and convert it to time series using
   task parallelism


   - PyAverager - reads data off NetCDF files and performs averages


PyReshaper has been used in conjunction with PyNIO, which is the Python
interface to NCL's file I/O capabilities.

PyReshaper/PyAverager description and download:
https://www2.cisl.ucar.edu/tdd/asap/parallel-python-tools-post-processing-climate-data

Their SEA2015 presentation:
https://sea.ucar.edu/sites/default/files/Mickelson-Bertini_SEA2015.pptx

A user seminar on task parallelism:
https://www2.cisl.ucar.edu/training/library/user-seminar-presentations

​We are very interested in talking to users like yourself who have use
cases of working with large data files, or lots of data files, or both.
Would you like us to contact you about this in the near future?

--Mary​

On Fri, Jun 26, 2015 at 3:42 PM, Zhifeng Yang <yangzf01 at gmail.com> wrote:

> Hi
>
> Since currently I am employing NCL to deal with tons of large data files,
> I am considering to parallelize my NCL code. I searched online about this
> and found that there were some experts from UCAR working on this project.
> However I am not sure what's the process of this project. Does anyone know
> about this and give me any resources on the parallel NCL? Thank you, Zhifeng
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20150702/f96dd879/attachment.html