Subject: Re: [nmr_sparky] Slow contour drawing
From: Thomas Goddard
Date: Oct 23, 2006

Previous: 250 Next: 255


Hi PP,

I appreciate your suggestions. I agree that the printing / image
saving, and wheel mouse zooming would improve Sparky usability and I
could add many more ideas.

It may be that much faster contour drawing could be done by just
changing a few parameters in Sparky code (contour tile size, default
cache size, default brick size used by pipe2ucsf, bruk2ucsf, ...)
and recompiling. If the bottleneck is the large number of contour
tiles this could make it 10 times faster. I will try that change
when time is available. We have no funding for Sparky development
though, and I have more electron microscopy projects than I can handle
so when time is available tends to be measured in months or years.

Im not clear from your emails why Sparky is so slow in your case.
Without specific numbers -- data size in grid points, how many contour
levels, are contour levels down in the noise, how many peak markers
and labels (thousands?), how many visible peaks -- I cannot be sure
what optimization would help you. As I mentioned, my own test case
was an uncommon way to use Sparky. My impression is still that you
may be using Sparky in a way that you are accustomed to working with
other sofware. But Sparky may have better ways to achieve the same
result that you are not yet aware of. Lacking details that is just
speculation.

Tom