Exporting from Pyprocessing to Povray

[monkstone]

I have recently experimented with creating a menger in pyprocessing, opengl rendering performance is not really that great, so experimented with exporting the geometry as a mesh2 object which produces pretty impressive results see my blog.

[Flavio Coelho]

Your Menger.py drawing is indeed heavy, with all that recursive drawing. However, running it under pypy 1.8 gave a somewhat better performance:

pypy: 1.17 fps

CPython: 0.84 fps

elegant_ball.py on the other hand, got much better framerate under pypy:

pypy: 13.76 fps

CPython: 3.89

Pyprocessing looks a lot better under pypy, doesn't it? 

If you don't know to get pypy running on your system, please see this.

cheers

On Thu, Apr 5, 2012 at 16:12, monkstone <mamb...@gmail.com> wrote:

I have recently experimented with creating a menger in pyprocessing, opengl rendering performance is not really that great, so experimented with exporting the geometry as a mesh2 object which produces pretty impressive results see my blog.

--
Flávio Codeço Coelho
================
+55(21) 3799-5567
Professor
Escola de Matemática Aplicada 
Fundação Getúlio Vargas
Rio de Janeiro - RJ
Brasil

[monkstone]

Thanks, I have since re-factored my code menger2.py to separate out Box (formerly Cube) and Mesh logic, which might also help a bit, I will certainly give pypy a go. However I know from work I did on a vanilla processing sketch that a bigger performance gain could be made by switching to using GLSL shaders for rendering see a discussion on the vanilla processing forum, however I don't feel up the technical challenge, but that is where vanilla processing is heading.

[monkstone]

I'm pretty impressed with pypy it did definitely does improve performance, what I did was install pypy (extracted binary to /opt and used update-alternatives to create a symbolic link to /usr/local/bin/pypy), and then I re-installed both pyglet and pyprocessing using pypy. Is there any advantage in bootstrap recompiling pypy, I'm using a 64 bit linux distro?

However I suspect greater performance improvement could be obtained by switching to using GLSL rendering as they are currently doing with vanilla processing albeit with some teething issues.