question

elizabeth.l avatar image
3 Likes"
elizabeth.l asked Matt Long answered

Performance comparison - 3D vs Process Flow models

I was wondering if you have any information that compares the performance of 3D models and process flow models. If there's no documented information, do you have a sense as to how the 2 compare, in terms of performance? Here are a couple simple examples of what I mean. Creating a label on a token vs an item (box). Creating an item via a Source vs creating & moving an item via the Create Object activity. In general would you say that process flow models run faster than 3d models?

FlexSim 16.2.2
process flowperformance
5 |100000

Up to 12 attachments (including images) can be used with a maximum of 23.8 MiB each and 47.7 MiB total.

Matt Long avatar image
4 Likes"
Matt Long answered

It is very difficult to compare speeds of 3D models vs Process Flow models. Even if you created two models, one using all 3D and one using Process Flow, that performed the exact same logic, the speed differences would not reflect the speed differences for every model.

Differences like creating a label on a token vs a flowitem will not have any impact on execution time. Creating a flowitem with a 3D source as opposed to a Create Object activity will also not have an impact on execution time.

Here are some general tidbits of information that may help in understanding when it may and may not be faster.

  • A flowitem requires more memory than a token. This also means it requires more computing power to create a flowitem than it does to create a token.
  • The more events that a model creates, the more time is spent executing those events. Depending on your requirements, a 3D model could have a large number of events, or Process Flow could have a large number of events.
  • Each Process Flow activity performs a simple task, assigning labels, starting an animation, acquiring a resource, etc. In order to piece together a complex set of logic, many different activities are required, possibly hundreds of activities. Though the execution time for an individual activity may be short, executing hundreds of activities can begin to add up.
  • Referring to the previous bullet, the same logic that took hundreds of Process Flow activities may be possible by writing a large chunk of code. In this case the model may have been able to eliminate the creation/destruction of tokens, writing and destroying excess labels and it does not need to collect statistical data (an activity updates its stats each time a token enters or exits it). In this case, the purely code logic will execute faster than in Process Flow.

Keep in mind that Process Flow wasn't designed to be faster than writing code or using other previous modeling techniques. Process Flow was designed to make the process of building a model faster, more logical and easier to understand and maintain.

The fastest simulation model possible would not use any 3D or Process Flow, but would instead be created entirely in code. Of course this would not be a very friendly simulation...

5 |100000

Up to 12 attachments (including images) can be used with a maximum of 23.8 MiB each and 47.7 MiB total.

Sam Stubbs avatar image
2 Likes"
Sam Stubbs answered Kathryn McNeal commented

As a general rule of thumb, yes they do. The number of calculations needed to animate and position objects in "3d space" and then update that image constantly to you is going to exceed the calculations a token is going to process when moving from activity to activity. Of course it depends on the activities from there.

· 1
5 |100000

Up to 12 attachments (including images) can be used with a maximum of 23.8 MiB each and 47.7 MiB total.

Kathryn McNeal avatar image Kathryn McNeal commented ·

Does this hold true even if you turn off "show content" for all of the objects in your model?

0 Likes 0 ·