After I posted about the jellyfish a couple third-party observers commented, paraphrasing, 'Duh, of course you need to turn the hidden stuff off'. (Quick review: The issue there was clustered arrangements of elements, where AC was taking a long time to sort out the hidden lines.) Maybe so, but the last time we visited this issue, which was probably way, way back when, it didn't make much difference. But our projects today are more complex, with many more polygons, so we should look into it.
I tested the front elevation, with fills, of the hairiest project I could find. On the Quad G5, generating the SE from scratch took 2:45. This is with all the interior elements on.
I made a new layer combination and hid the interior elements. This includes:
• Structural columns, beams, footings. (Remember, we don't do joists or rafters.) (Also remember, architectural columns are layered as walls.)
• Interior walls (which shuts off the doors, natch)
• Interior trimwork
• Appliances, mechanical equipment, plumbing fixtures
• Stair parts
• Fireplace flues
Generation time: 2:15 (18% better) Noticeable, if not life-changing, improvement.
The question: Should the templates be equipped with such an LC?
• The improvement would only be noticeable on very complex models.
• A separate LC would mean a separate clone, making the view map more complex.
• SE generation will improve as we move to Intel hardware.
The dramatic slowdown in the jellyfish was caused by 'clumping' of elements in an unusual case. Such cases could be handled piecemeal, not burdening the template. Templates should be geared toward standards and typical usage.
I'm going to say the improvement isn't worth making the templates more complex. On a complex model, it might be worth forking the A2 LCs.
On a truly large, multistory building, I think it would be advisable. As you add lots of stories, it would seem you're adding interior polygons faster than exterior.
And I reserve the right to change my mind after another 8 years.