2018 Large Assembly Performance Community Initiative (insert,move/align, joints)

[nathan.skalsky]

I'd like to propose an initiative centered around improving the performance of large assemblies in Fusion 360 and coming with a realistic specification of performance expectations for large models (ie. how many joints/faces, etc. can a model have and expect to be performant).  

We can use this idea submission to gauge community interest, something the Autodesk team can use to ensure this is weighted appropriately, as we all know, work on this means less development bandwidth for other things, new features,etc. 

This should be primarily focused on areas that directly affect real-time interactions such as insert/XREF, joints & positioning and the associated data structures and not GPU rendering or ray-tracing.   This covers commands such as move, align, create joint, rigid group. 

I think there is significant community motivation to help, likely primarily by curating a small set of representative models that Autodesk can use, as well as providing feedback on improvements.   

Thoughts?

[Anonymous]

Yes!  We want more Marble Machine X in Fusion!!!

It has been great learning Fusion 360 and watching the Wednesday updates along the way.  It would be a shame if the whole CAD had to be shifted to Inventor.

[wintergatan]

Thats an idea i can get behind! 🙂

[kb9ydn]

Well now this is quite interesting!  Large assembly performance is one of my main gripes with Fusion right now.  In fact it's one of the main reasons I continue to use Solidworks instead of Fusion for all of my design work.

But before I go into what I think one of the main issues impacting performance is, I'm wondering if it might be better to start a thread about this in the Support or Design sections of the main Fusion forum?  I think this topic needs a fair amount of discussion and (IMO) it would be better to do it there instead of here in the ideas forum.  The ideas forum works better with specific well defined ideas for improvements to Fusion, and not as well for topics that are less well defined or that need more discussion.  It will most likely get a lot more visibility there too.

After the specifics are worked out, they could be posted back here for voting.

C|

[Anonymous]

Is there any work being done to improve large assembly performance? I have 

a fairly large project coming up, and don't want to start it with a software that will not be able to finish the job.

[wintergatan]

---

Hey William, I do not know if it is helpful but the marble machine x I am working on has 8000 parts now and I managed to make fusion 360 handle it in the end. With workarounds. You have to divide your project into many subassemblies with correct relation to the origin point. And not use joints. So I would say with workarounds, fusion 360 can do it. You have to design inside smaller subassemblies, you can not design in the big assemblies but after getting used to the workflow of upating designs it feels natural. But with a big disclaimer that I do not know how massive your project would be. If you need help I would be willing to give comments. all best/ Martin / Wintergatan

---

[Anonymous]

Thank you for responding.
My project will likely only be about 5ooo parts.

Are you able to use joints in the sub-assemblies, or none at all?

[wintergatan]

I have forbidden the use of joints in the entire project, (except rigid group) I have developed a workflow that doesn't need any joints. Most often when there is problems is because of the joints. So if you really want to use joints a lot. I wouldn't go with fusion 360 on a project with 5000 parts. Seems like inventor does better with joints. On the other hand - there is nothing I want to do in fusion right now that I need joints for. If yo want to make nice animations joints would be nice to use. I hope this is helpful somehow.

[dean.dekovic_gideon.brothers]

Hi,
This idea is so important that making 150 Autodesk accounts and liking this idea seems like a good idea.

The graphics detail can also be heavy and seems unnecessary, for example seeing a car in a curved surface on a bolt while zooming in/out.  

[michael.woodrum]

I found this while, again, attempting to reduce the time it takes to update things within a large assembly. I hope that the workflow in the future is less laborious. Maybe Autodesk will get this figured out in the future. I hope so, otherwise the software may go away. I was thinking that Fusion will basically replace most products from autodesk in the future. Maybe not replace, but combine all major functionality from all the products in the line.

[Anonymous]

The first place to start is add a model performance tool to the inspect toolbar so one can see where the bottlenecks are located in ones model. Report this data to A360 so management can see. Then aggregate all this data via a dashboard for the Fusion 360 development team so they can quickly see the performance pain points of the users and allocate development accordingly.

Developers, please leverage my GPU!