Sculpting and relations to gpu and cpu

Recently I was fortunate enough to upgrade from intel 10400 to 10700.
I thought blender in general would get faster and it did but not sculpting.

much later I upgraded from nvidia 1660 Super to 3060 ti and man oh man sculpting is much smoother.
I no longer need to hide meshes in matcap view just to be able to do smooth strokes and such. (23.8 m tri)

So my question is, as far as I know, most sculpting codes are c codes that runs on cpu I presume? So is there a bottleneck in viewport that causes it to be this gpu dependent? I tried turning off anything on overlay and get as simple as I could and have no modifiers but multires. and most of them almost did nothing in terms of performance gain.
Am I missing something?

If you search for benchmark 10400 vs 10700 then the overall speedup is… not soooo much. Passmark
12500 vs 17500 (single 2600 vs 2900); this is a max of 40% (theoretically) ans as you mentioned not really noticable. But the main part in sculpting is… showing/displaying the millions of polygons, so it make sense that the major speedup is by a graca which gets an effective speedup 85% and average score at 79% higher… (again searched benchmark 1660 Super vs 3060 ti on you favorate search engine).

that might be it, although both gpu’s barley worked in sculpting.
also isn’t sculpting multi threaded? shouldn’t that help?

this was so unexpected.