I would like to inquire about how to transfer data through Speckle, like if I select maybe 15000 element in Revit or Navisworks, it will take too much time to analyze and send especially in Navisworks.
So is there any way to do it faster?
Could you define ”too much“?
We try very hard to improve performance and periodically do improve things. We are also testing prior conversion speeds to check for regressions.
We’d love things to be instant but sometimes the internal API is the bottleneck - similar to dynamo having a less than stellar reputation.
Best practices do tend toward not sending everything at once, but that’s not a fix all either.
I mean, sometimes it takes more than 1 to 2 hours to transfer a part of mega project.
Specifically for MEP models, if you can suggest something to do, I’ll appreciate that.
I recommend we can develop custom to do that :
- Send with geometry include
- Send just with metadata
- Send just with geometry
- Send by Item type and merge by branch.
We have the same discussion internally from time to time. We also made a post about it here for speed comparisons with Rhino/GH, because a few colleagues were unhappy about the speed compared to Rhino.Inside.
Performance Checks Rhino/GH/Revit - Features & Ideas - Speckle Community
In the end, the time factor will be the most crucial one for the adoption of Speckle in order to succeed. Example: If sending Geometry from Grasshopper to Revit is drastically slower than Rhino.Inside, people will prefer Rhino.Inside … even if Speckle has in the overall process more benefits. If the overall file-based process is way shitier, Speckle will prevail. The same applies for Navisworks and Speckle vs. the nwc converter. Maybe some processes will only work file-based due to API limitations … but this is out of our hands. We cannot expect for a Trabi to drive like a Porsche … Trabi will be Trabi (aka Autodesk Revit API … and other vendors).
There are still some things the Specklers are able to optimize, but this needs time … as it is a big undertaking . For the platform to work on bigger projects, we need to have some patience. This platform has evolved drastically over the last years, therefore I’m fairly confident that they will manage to tune the gears a bit .
As a former BIM Coordinator, I know the pain; I have spent hours exporting NWCs from MEP Revit models.
That fact alone means I know full well what a pain it is and that the underlying tech isn’t inherently fast. We can investigate, as @AlexHofbeck suggests, on a case-by-case basis if there are generic solutions we can optimise for or if specific cases are a pinch point.
If you want to share the model taking 2 hours, I am happy to take a look if it is generally slow.
From the Navisworks connector POV, I have a couple of places where I have prioritised robustness over performance, but I have some avenues to pursue. But it’s getting better; since the first Alpha release in December '22 til today, some models that were taking those sorts of times for Send are now in the order of 10 minutes - Navis users have some really big models.