[ncl-talk] gsn_csm_contour_map in Panel plot runs out of memory

[Mary Haley]

Hi Melissa,

We need to know more about the problem in order to determine why so much
memory is being requested.

As Barry asked, did you get an error message?  In particular, did you get
an error message like this:

 fatal:ContourPlotDraw: Workspace reallocation would exceed maximum size

and then add some code that looks like this to deal with it?

  setvalues NhlGetWorkspaceObjectId()
    "wsMaximumSize" : xxxxxx
  end setvalues

If so, how big did you set wsMaximumSize to?

Also, how big is the variable that is being passed to gsn_csm_contour_map?
You can do a "printVarSummary" on it and then post that back to ncl-talk.
How many contour levels are you generating for each plot? This can cause
more memory to be required.

If these are large arrays that you'r contouring, then using raster fill
might help speed things up:

  res at cnFillMode = "RasterFill"

These are just a few of the suggestions.  You may need to review the
variables that you are reading and/or calculating to see how big they are,
and whether you really need to do this kind of calculation across the whole
domain.

As a final step, it would help if we could look at your script, and even
better if we could try running it on casper.

Thanks,

--Mary


On Tue, Nov 13, 2018 at 11:00 AM, MELISSA KAZEMI RAD <
mk1369 at scarletmail.rutgers.edu> wrote:

> Hi,
>
> I am trying to create panel plots, (2 contour and 4 xy plots) using
> "gsn_csm_contour_map" and "gsn_csm_xy" from wrf outputs for different
> times. However, the job runs out of memory for one of the contour plots,
> only for certain times. (The resulting figure only lacks one of the contour
> plots!)
>
> I am running ncl scripts on NCAR's Casper, and am using a total of 1500 GB
> of memory! I was wondering if there's a problem with gsn_csm_contour_map,
> and if not, if there are ways to make the script more efficient. The
> contour plots and xy plots are drawn separately using gsn_panel.
>
> I tried plotting the contour plot separately for those certain times, but
> it still runs out of memory. I have encountered the same problem for some
> other variables before.
>
> I really appreciate some help with this issue.
> Melissa
>
> --
> *Melissa Kazemi Rad*
> *Atmospheric Sciences Department*
> *Rutgers University*
>
> _______________________________________________
> 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/20181113/2146b43a/attachment.html>