Hello,
Perhaps it's worth pointing out a couple of reasons why Ansig is faster
than Analysis.
(1) Ansig only uses pre-calculated contours. This absolutely definitely
speeds things up. Analysis now has this as an option. Ansig also has a
limit (8??) on the number of spectra which can be drawn.
(2) Ansig is written just in Fortran/C and OpenGL. There is no beating
this combination for speed. Analysis uses Python and Tk to wrap around
the C and OpenGL (or Tk) drawing code. Even the C code in Analysis will
not beat the Fortran/C code in Ansig because the latter is special-tuned
for OpenGL, and we have a Tk option. And in the Python world, somehow Tk
does not behave very optimally. Part of it is that too many refreshes are
happening. But I think it goes beyond that. The Tk event loop just
seems to be slow. And we also have lots of Python objects, many of which
have to be coordinated with the C world, and this is bound to slow things
down.
I am pretty sure we will be able to speed things up with some effort, but
how much improvement we will get and when this will happen I'm not sure.
Wayne
|