Thanks Michael.
I guess you are right that extra time is due to overlapping the ghost
layers.
If you want more performance, you should use .pvti and .vti instead .pvtr
and .vtr to store rectilinear data.
My grid is not necessarily uniform even in one direction. If I remember
correctly, *.vti file
Hi Seb,
Did you try to specify the WholeExtent to be the Whole system extent
and not the piece one ?
Yes I did, but it does not seem to be correct. Some wiggles appear on the
contour of the first partition of the dataset.
In both *.vtr files I changed the line as:
But no luck!
Do you think the
Hi Mohamad,
I thought you were working on an unstructured grid. As it is not the
case you shouldn't bother with the array that I talked earlier.
But you should provide the information that each piece is a part of a
whole system and not a set of whole system shown in the same 3d world.
Did you try
Hi Seb,
I did not define any separate array. The only thing that I have done when
writing the *.vtr file for each subblock is something similar to this
File0.vtr
**
Hi Michael,
I guess *.pvtr file solved my problem. However, I have noticed combining all
my *.vtr files via the *.pvrt file, then plotting a contours, it is slower
than just loading it via *.pvd file.
Thanks for the tip though.
What is a .pvd file?
To my knowledge, it is a very useful and powerfu
Hi Mohamad,
Did you add a data array "GhostLevels" to specify for each cell its
ghost level ?
It's an unsigned char data array where the values are the depth into
the next block for of each cell.
By that I mean:
- 0 is the cells of the current block
- 1 is the set of cells that share points in the