[pdal] Python and chipper filter

Lars laasunde at hotmail.com
Sun Sep 16 23:39:02 PDT 2018


Andrew,

Appreciate the explanation.

It would be useful to understand why this issue occurred. Does appear to be reproducible 😊

Regards, Lars

________________________________
Fra: Andrew Bell <andrew.bell.ia at gmail.com>
Sendt: onsdag 12. september 2018 15.41
Til: Lars
Kopi: pdal at lists.osgeo.org
Emne: Re: [pdal] Python and chipper filter

The chipper filter creates sets of points for each tile.  These sets (called point views) pass through the pipeline individually and are processed separately by each downstream filter.  In your case, the chipper should create two or three point views (since the count limit on the points is 10,000 and the capacity limit on the chipper is 5000 points).  This essentially means that the python filter gets invoked two (or three) times.  When you reverse the order of the python filter and the chipper, the python filter is only invoked once, since at that point in the processing, only one point view exists.

All that said, the differences you're seeing seem extreme and I think it would take more investigation to figure out the root cause of the timing differences.

On Mon, Sep 10, 2018 at 3:47 AM Lars <laasunde at hotmail.com<mailto:laasunde at hotmail.com>> wrote:
Hello,

We have created a pipeline with a python script but are seeing some unexpected behaviour. The python script currently does nothing to exclude the script as culprit.

Running the script with 10k count and pipeline is completed in less than a seconds. Increasing to 50k count and it takes to 2 seconds but increasing to 80k count suddenly it takes 150 seconds!!! Attempting to use 90k count and after 25 minutes it is still not finished (pdal is constantly using 12,5% --> one core).

If we changed the order of the filters (python first and then chipper), then 10k count took less then a second and 100k takes 4 seconds.

If we remove the python filter then 10k count took less then a second and 100k takes 4 seconds.

This appears to be related to the python filter.

Why is the the attached pipeline so slow? Is this a software issue or poorly designed pipeline?

Using PDAL 1.7.2 on Windows 10.

kind regards, Lars
_______________________________________________
pdal mailing list
pdal at lists.osgeo.org<mailto:pdal at lists.osgeo.org>
https://lists.osgeo.org/mailman/listinfo/pdal


--
Andrew Bell
andrew.bell.ia at gmail.com<mailto:andrew.bell.ia at gmail.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/pdal/attachments/20180917/dfd5d0ef/attachment-0001.html>


More information about the pdal mailing list