Thanks for running DSS again - it is a pretty good benchmark for placing stress on your CPU and I/O / RAM components.
The registering process for a field of stars looks to identify all the brightest stars on the X-Y grid - and see between each shot how much if at all each is moving across a frame - then aligning all the frames to the best match acording to which shot it thinks has the best signal to noise. That is the easy bit. The stacking is very CPU intensive - as its not so much the stars but all the glowing gases between them it is trying to correct add (across all colour channels) to create a master 16 or 32 bit TIF file. Once you have this you may typically choose to:
- Align all the colour channels - so at a guess assume there is as much Red, Blue and Green in each shot (generally a bad guess - but great for learning from) - so move each colur peak distribution curve until they overlap.
- Play with the Dark, Dim, fait and bright zones of you combined shots - so here you are moving the light curve so it only starts to detect once you are partly into each colur curves' distribution curve - basically throwing out the crap signal that is too faint to bother with, boost the middle and reduce the brightest areas
- Add generally 15 - 22% saturation in case all your colour channels are too washed out
- Now once you press process it can take some time - generally about 3 minute 20 seconds on my rig for 30 light frames - from memory my kid's I5 PC did this in around 2 minutes 30 seconds - so not a huge leap. Interested what more cores does to the timeframe - I guess it scales very linearly as I would say its dividing all the photo X-Y grid into smaller rectangles and throwing each to a different core.
I still have one more PC (the office PC) to upgrade - so next year it might get a Threadripper to contrast Intel to AMD.