Hi,
This time i work on a big project and when i try to reaload a adaptive family with many “windows” revit take his time and can load during 40min to 1h30.
My family is very heavy (251 Mo) but i optimized it. (357 Mo before)
Perhaps, when i look at the task manager revit only use 10% of CPU and 6 Go of RAM and, sometimes, 1% of GPU
I have a Xeon® W-2133, 32 Go of RAM and a P4000 8Go
Why revit isn’t use a minimal of, i don’t know … 50% of CPU ?
Hi, most of the processes in Revit are done on single core. You can check this post on Autodesk forum for more info:
https://forums.autodesk.com/t5/revit-ideas/revit-using-multi-thread-on-multicore-processor/idi-p/7356716
also you can check which functions are Multi-threaded here ( Revit 2017):
I allready see this but, i see it’s Revit the problem in fact …
Did we have to take a processor with a high frequency to be more powerfull in revit ?
If you take a 4GHz to change a 3.5GHz did you think that can be a solution to “accelerate” revit work ?
It might be a solution. I started working with more Revit instances open (because im working on partial models), then I just assign instances to different threads. This is not an option on Federated model or Single server files, since you cant open file more than one time.
I hope that new processor does the trick for you.
To make a solution on this post :
For the moment we can imagine that a processor with a high frequency can be more powerfull for revit.
This isn’t an official solution.
Officialy, autodesk just say to optimize your model and don’t put many DWG in your revit project.
I let you search on Autodesk website different post about optimization model (it’s simple you will see).
A more powerful single core processor will outperform many operations in Revit, but not all. Typically you can achieve faster calculations by reducing scope. Load the family in as a new family (2 seconds of calculation time), and replace in chunks rather then rebuilding all n panels at once. Had a job once with like 20k curtain panel family instances which had similar issues due to RAM and virtual memory constraints - total reload time was 2 hours. Replacing in groups of 1k reduced total time to like 20 minutes + 20 extra sets of clicks (minimized by selection sets created via Dynamo). Might not work for you but it could be worth a try -just be careful as adaptive components can be uncooperative in this type of situation.
The reason stuff like loading an adaptive family into the project is single threaded becomes clear pretty quickly if you think it over. If the first adaptive component hosts points on the second and third, and the 4th based on points in 2 and 3, how can Revit know where to put panel 4 before it has placed panel 1?
Optimizing your model is a key aspect in performance, including stuff like removing warnings (if the adaptive component was room bounding above and the room had an unrelated slightly off axis error the calculations take longer at the placement of each panel), not being overly aggressive with constraints (over constraining is a pretty obvious error, but if we had hosted the panels above to a curtain wall or repeater surface it would load faster due to less prerequisites), and keeping geometry simple (if the adaptive component was a cube, couldn’t we just use a standard curtain wall?). Management of these types of issues is really as much of an art form as it is a science.