What I have noticed is that if I use one main layer and a number of sub layers instead of putting parts on their own layers, I get better performance.

I am not sure about Viacad as I use Shark, there is a compact option in Shark under the main file menu, I use this when I am sure that my model is complete.
Compact removes all history and gets rid of deleted parts which is a must because Shark will hold onto deleted objects somewhere in its memory.

Try this experiment Larry, use a single part that has a number of features and set it to super fine, then render the part a number of times and watch your ram climb.
Here's the really annoying thing, once the has reached about 60% close that file but leave Shark open, you will notice the ram does not drop off, it will stay at 60% and any new files will start off at 60%!
Blender a free program has absolutely no problem, sure the ram climbs but once you start a new file without closing the program, the ram goes back to where it should be.

I submitted this a year ago and they still have no idea what to do about it, Shark is the worst program I have come across for resources usage.
You can have a computer worth $100 000 and Shark will still perform the same as it does on a $300 laptop.

The day that these marketing jokers realise that customers would much rather have stability over new features is the day the sun will go supernova.
All this is driven by marketing people who do not have a clue and that's why their incompetence is so blatant now.
These are their end days and they are scraping the bottom of the barrel to pay for the luxury lifestyles that they cannot earn.
They have been riding on the backs of skilled people like Tim and ourselves for so long now they think that it is their God given right to fleece customers.

Word to the wise for the marketing department, start at the bottom and stabilise any product before selling it to customers.
Because if you don't you will have no customers left and there are no more products to rip the guts out of while you pretend to be superstars with your expense accounts and luxury company cars.

That's what compact does, it removes all the history and purges deleated items, compact your files on a regular basis for improved performance.

I have found that individual part modelling is far more productive than assembly modelling, if I have a large assembly I will work on individual files and then import them into a final assembly.
If my design contains a relatively small amount of components I will do it as an assembly, but for large assemblies I do individual files.

My understanding is that parametric modelling requires an active history tree, direct modelling does not, Shark has both but both require a history tree in Shark.

Have you used the Remove Links function that can be found under the edit menu?

If have two cubes and two spheres, place the spheres half way on top of the cubes so they are joined.
Then use the subtract solid command to subtract each sphere from their cubes, keep the control key pressed when you subtract the spheres.
What you end up with is two cubes with each sphere subtracted from them, if you move both of the spheres the hole in the cube fills up.

If you drag select one of the pairs and use the Remove Links command you will see that pair has lost its history, when the sphere is moved away the impression in the cube does not fill up.
Whereas on the other pair that was not selected when the Remove Links command was used still retain their history and the cube will close the impression of the sphere when the sphere is moved away from the cube.

Edited by user Sunday, January 31, 2016 10:41:49 AM(UTC)  | Reason: Not specified