Re: Slow Irregular3DSet-Constructor

• To: visad-list@xxxxxxxxxxxxx
• Subject: Re: Slow Irregular3DSet-Constructor
• From: Curtis Rueden <curtis@xxxxxxxxxxxxx>
• Date: Fri, 15 Sep 2000 11:31:20 -0500

>> I'm new with VisAD and not jet very experienced with its mechanisms. I'm
>> trying to write an Application for presenting the measured data of a
>> satellite in 3D. Therefore I constructed an Irregular3DSet. I have to
>> handle with a lot of data. So I tried to construct a Irregular3DSet with
>> nearly 5000 datapoints. On a Pentium II-400 System it took about half an
>> hour to run the Irregular3DSet-Constructor.
>>
>> My questions are:
>> - Is it normal that the Irregular3DSet-Constructor is that kind of slow?
>>
>> - Is there another possibility to visualize my data?
>
>The Irregular3DSet constructor is slow because it invokes an
>algorithmn for computing the Delaunay tetrahedralization of
>your 5000 3-D points, and even the fastest Delauany algorithms
>are slow for this task.

I agree with Bill's advice to use Linear3DSet or Gridded3DSet if possible.
However, 5000 datapoints should not take half an hour to compute on a
P2-400.  Tetrahedralizing 5000 data points with the Clarkson algorithm
takes less than 30 seconds on my P2-400 system.

Please run "java visad.Delaunay 3 5000 1 1" to test the Delaunay
tetrahedralization algorithm.  It will report back to you the amount of
time needed to construct the tetrahedralization.  If the tetrahedralization
finishes in a reasonable amount of time (i.e., less than 30 seconds), then
there could be a problem with your particular data set, or there could be a
bug elsewhere in VisAD causing the slowdown.

-Curtis

• References:
  • Re: Slow Irregular3DSet-Constructor
    • From: Bill Hibbard