[Paraview] Calculator performing poorly in ParaView-3-8 branch

[John Patchett]

I've noticed the calculator has started performing substantially
slower in the ParaView-3-8 branch when taking the magnitude of a
vector as compared to the Head before the branch.
I've attached a graph summarizing the performance degradation.  I used
the Mandelbrot source and took the magnitude of the coords vector to
collect data for the graph.
One of my machines averages nearly a 6x slowdown and the other a 4.5x slowdown.

The *-march is a Head build  from 3/23/2010.
3-8 is the branch as of yesterday afternoon.

I (believe) the change that affected the calculator performance
happened after the branch as I noticed it after I updated the branch (
I wouldn't bet too much on this though)...  I don't know if the Head
is affected.

Thanks,
--John Patchett.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: calculator.eps
Type: application/postscript
Size: 17802 bytes
Desc: not available
URL: <http://www.paraview.org/pipermail/paraview/attachments/20100413/9c078af8/attachment-0001.eps>