Reduce accuracy to reduce memory and time cost

Hello

Is it possible to reduce the accuracy of grasshopper in order to reduce the amount of time and memory a system uses.

I've made a system that scans a model of a site 200 x 200 metres. It simply pushes a array of lines through the model and finds the intersecting points. The tolerance grasshopper seems to be working to is about 0.00001 metres. For what I am doing 0.01 metres is fine.

Is possible (and would it save time and memory) to reduce the accuracy in grasshopper?

Replies to This Discussion

Permalink Reply by Daniel González Abalde on October 4, 2016 at 5:38am

Grasshopper uses the tolerance of active rhino document. So, in Rhino go to:

Tools > Options > DocumentProperties.Units

Permalink Reply by John Everist on October 4, 2016 at 5:53am

Thank you Daniel

Permalink Reply by David Rutten on October 4, 2016 at 6:12am

It might not save as much time and/or memory as you might think. It will for certain operations but not for others. To see why, you must understand how tolerance settings are used. There are two broad categories:

Determining coincidence or tangency or perpendicularity.
Fitting.

When you intersect two curves, Rhino calculates the most accurate solution it can, regardless of tolerance settings. Two curves in 3D almost never exactly intersect, they only tend to approach each other before diverging again. So once Rhino has found all these areas where the curves approach each other, it will then decide which count as genuine intersections based on the current tolerance settings. And it will also decide which ones count as point-intersections and which ones should be overlap intersections. In cases such as intersections you will save little memory and processor cycles by switching to lower accuracy.

However when you intersect two surfaces the other kind comes into play. Rhino will generate a boatload of intersection points in the UV space of both surfaces, and it will then attempt to fit a curve through those points. If a very high accuracy is required this fitted curve will have lots and lots of control-points. That will consume more time and more memory, especially if that curve then goes on to play a role in other operations.

ps. I've painted a simplified picture here, there are sometimes optimizations that will take tolerances into account in order to reduce computational overhead.

RSS