Generative Design process measurable?

, ,
1

When applying the Generative Design process, you must of course also take the processing time into account. However, is this measurable anywhere?

Fig. 1
Fig. 11264×658 50.4 KB

Fig.1 (picture of AU session “Generative Design At Hogwarts: Using Tech Instead of Magic”)

Thanks in advance.

2

In terms of an initial estimate, you can:

  1. Use the TuneUp view extension to get the run time for the graph.
  2. Divide that by 6 because you’ve got six cores working concurrently with Generative Design.
  3. Take the generation and population settings of your graph and multiply them, so a 100x100 study is 10000.
  4. Multiply the result of step two and step three - this is about how long it will take.

Obviously this is an estimate - some combinations of inputs will run faster than others, the tuneup test doesn’t take into account recording the results or the geometry, and of your graph accesses data outside of the dyn itself that may reduce the capacity of the six cores.

1 Like
3

@jacob.small thanks for the input (results see fig. 1)
However, additional questions:

  1. Are there any tricks to include the geometry in the calculation?
  2. How do you deal with the “randomize” type in the calculation (generations multiply … ) ?

Fig. 1
Fig. 11461×478 95.5 KB

Fig. 1

Thanks in advance.

4

  1. Geometry will impact graph runtime. Higher complexity geometry will result in slower runtimes. Generally, stay left in the geometry hierarchy to reduce complexity (chart is 3/4 of the way down this page): Geometry Overview | The Dynamo Primer

  2. Randomize is basically on generation of optimize. Same with cross product (use the total number of tests), and like this.

2 Likes