Pycam performance

Post Reply
brianpaavo
Posts: 3
Joined: Thu Jul 26, 2012 12:35 pm

Pycam performance

Post by brianpaavo » Fri Jul 27, 2012 5:50 am

I have a relatively simple object (you can grab .stl at http://www.wormguy.com/PanelWindow10mmAscii.stl it's about 80x70x10 mm) , but since it has radiused corners I expected it to have a lot of triangles. There ended up being over 6000. When I tried to generate a tool path in Pycam I thought my weaker Linux machine might have stalled so I used my WINXP box instead (SP3, 2.80GHz Duo with 3.5 GB Ram). After nearly 24 hours of processing and completing at least 10 of the 11 layers Pycam crashed with the usual lack of helpful information from windows. I was using a cylindrical cutter of 2.5 mm with 0% overlap 1.0mm max steps in a follow contour strategy, rough toolpath only.

Is this typical "performance" or must I have something terribly wrong? Is Cam Bam better?

viralata
Posts: 19
Joined: Tue Dec 04, 2012 9:26 am
Location: Cordelle-France
Contact:

Re: Pycam performance

Post by viralata » Mon Feb 11, 2013 7:45 pm

I have the same problem with pyCam on Linux: for a 5000 tris model, I can see the ram used growing upt to more than 20 gigs, so the computer is swapping a lot and become unusable, but at least it finishes the job. But 10 hours for a simple model is really a big problem. For 2D (from svg) it's working nicely though
http://viralata.fr : some architecture and blender, mostly in french

viralata
Posts: 19
Joined: Tue Dec 04, 2012 9:26 am
Location: Cordelle-France
Contact:

Re: Pycam performance

Post by viralata » Mon Feb 11, 2013 9:36 pm

After some reading, I saw that a new version 0.6 was supposed to be out some years ago, but the developpement doesn't seem to be completly dead after that:
http://fab.senselab.org/en/pycam
At least it's less than a year...
So I tried the developpment version as explained here:
http://sourceforge.net/apps/mediawiki/p ... nt_version
I don't have anymore default tools and process and I had to create them, but what a huge improvment: for the same model, only 500megs instead of 20 gigs of ram, and half an hour of processing instead of 10 hours :D
Now it is usable !

I hope the development will continue
http://viralata.fr : some architecture and blender, mostly in french

Post Reply