Revit 2016 | Project Performance - Part 6

As I mentioned the other day in the conversation with Bill Debevc,one of the enhancements in Revit 2016 is improved management of System Family geometry. Multiple instances of editable families have always shared their geometric definitions, leading to much more efficient use of memory, but the same was not true of System Families, for a variety of technical reasons and history that were were able to surmount in Revit 2016.

The video below compares Revit 2015 on the left with Revit 2016 on the right, performing the same sequence of view operations on multiple instances of an piping family. Revit 2016 is done before Revit 2015 really gets going.

 

In this next video, a similar comparison of drawing pipes, Revit 2015 on the left, Revit 2016 on the right. Again, Revit 2016 finishes the task before Revit 2015 can complete its first operation.

 

Shared geometry allows us to leverage the GPU far more heavily than before, taking advantage of hardware that's become much more commonly available than when Revit was first designed and released.We're continually looking at early decisions in Revit and rethinking them in light of the demands of larger and more complex models and the improved hardware on the market.

Comments

[Armando Martinez]

Thank you guys for the post. In doing this comparison myself on my end, drawing pipes was nearly identical in both releases, nothing close to the lag shown in your "Drawing Pipes" video. Perhaps I have a later service pack (Update release 7) on 2015 that addressed this?

[Dourevit]

That video of 2015 is very fishy. If Revit MEP were THAT slow, we would definitely not be using it on large projects, especially considering that dataset seems very simple.

[Anthony Hauck]

The key differentiator here is the System Type’s Calculation setting (Performance vs. Volume Only), which was introduced in 2015 Release 2 (though, we also included in UR releases for all given the impact). The system has 4000 sprinklers, 8428 segments, and 4428 fittings. The system is a fully connected, though many users will split systems by physically disconnecting to mitigate performance bottlenecks. Our goal is eliminate that need. Hopefully that clarifies, and helps ensure apples to apples comparison.