Three comments:

a) I do not thing the "denoising" is cumulative.  I did a test "100 iteration" test, turning on denoise at iteration 1, and then another run turning it on at iteration 99.  I could not detect any difference.  I've thought about using some photo editing software and doing a "difference" between the two frames to highlight what I'm missing, but to my eyes the results looked the same.

b) Running denoise does use some time and resources, but very small, as in a fraction of a second.  There is a small hit when it "loads" the denoise capability, but after that the cost of turning it on early is minor.

c) For animation, you will need longer, not shorter, rendering times.  Why?  The noise you see in each frame will be subtly different, and what might not be noticeable on a single, static image will appear as flickering and shimmering on animation.  So that means for animation you will possibly want to render longer, and of course you'll be rendering many, many more images.

But... I'm still looking forward to seeing what we can do with it.

---

iClone 7... Character Creator... Substance Designer/Painter... Blender... Audacity...
Desktop (homebuilt) - Windows 10, Ryzen 9 3900x CPU, GTX 1080 GPU (8GB), 32GB RAM, Asus X570 Pro motherboard, 2TB SSD, terabytes of disk space, dual  monitors.
Laptop - Windows 10, MSI GS63VR STEALTH-252, 16GB RAM, GTX 1060 (6GB), 256GB SSD and 1TB HDD

Yes, I've noticed the dots as well. But most if not all aren't apparent when you save out an image file.
The only real 'critical' subject matter I've discovered so far is 'eyes' - which seem to need a lot more iterations to look convincing (especially on closeups).
After reading this post of yours here - I tried a much lower value of 30 iterations and was surprised that it CAN be OK on some shots.

--------------------------------------------------------------------------------------------------------------------------------------------------------

Please be patient with me ..... I don't always 'get it' the first time 'round - not even the 2nd time! :(  - yikes! ... 

● MSI GT72VR Laptop, i7 7700HQ 4-Core 3.8 GHz 16GB RAM; Nvidia 1070, 8GB Vram ● iClone-7.93 ● 3DXChange Pipeline 7.81 ● CC-3 Pipeline 3.44 ● Live Face ● HeadShot ● Brekel Pro-Body ● Popcorn FX ● iRAY ● Kinect V2 ● DaVinci Resolve17 ● Mixcraft 8.1

I was curious what you guys felt were the best setting for animations in the future (when we can render IRAY in ICLONE7). IRAY renders can take quite a while and if you have the iterations too low it will look like there are dots everywhere.
My idea for quickest results would be to have low iterations about 10-20 followed by a few frames with De-Noise turned on. So an example would be have de-noise set to come on after 15 iterations for the last 5 making it 20 iterations total.
I know very little about this so maybe someone who understands it better may have a better idea.