Question

My program is a visualizer that is constantly updating the display. I am trying to optimize it to increase the framerate, and my largest cost is gtk._gtk.main:

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    9.159    9.159   45.476   45.476 {gtk._gtk.main}
      868    8.146    0.009    9.884    0.011 xyz.py:291(gfx_transform_queue)
      868    7.705    0.009   12.657    0.015 xyz.py:322(gfx_draw_queue)
      868    3.886    0.004    7.755    0.009 xyz.py:240(gfx_queue_atoms)
   868000    2.638    0.000    4.952    0.000 xyz.py:365(gfx_draw_circle)
  7027010    2.111    0.000    2.111    0.000 xyz.py:314(cmp_queue)
  1736000    2.015    0.000    2.015    0.000 {method 'draw_arc' of 'gtk.gdk.Drawable' objects}
   869844    1.728    0.000    1.728    0.000 {numpy.core._dotblas.dot}
      868    1.635    0.002    3.746    0.004 {sorted}
   869872    1.261    0.000    1.261    0.000 {numpy.core.multiarray.array}
      933    1.100    0.001    1.100    0.001 {method 'get_pointer' of 'gtk.gdk.Window' objects}

Is this reasonable behavior, or is it possible I am doing something wrong to make gtk.main suck up a lot of time?

Answer 1

You’re interpreting the profiling data wrong.
gtk.main() is really just running your program as the user interacts with it. What you’re seeing is that it takes time to run your program, not gtk.main() itself.