It shows how slow your CPU's single threaded performance is and the trade off between latency and throughout in modern computing more than anything.
My experience from actually writing low latency schedulers in user space as well as the publicly available material - like in Ardour - suggests different conclusions from yours.
Keep in mind that a naive benchmark like "cpu usage" is entirely meaningless. What you look at is round trip latency required for a threshold of underruns/missed deadlines. Threading requires additional latency, and process synchronization even more. While I'm sure you report fewer underruns when splitting off into sandboxed plugins I'm suspicious if it's hitting the same performance as doubling or tripling the buffer size in terms of latency in the first place.
I am using i9 9900k @5GHz so I think it is quite fast. The plugins I use are CPU heavy and you can run limited number of them on a single core. I am not interested in low latency - I am running 1024 buffer, however I would like my projects to play smoothly even if there is latency. Ableton unfortunately does not work well with such use case as it won't parallelise where it could and sandboxing does just that and I can run more plugins, even if it seems less efficient.
For other readers: It is a misconception that sandboxing per-se enables parallelism. On the contrary, it only hurts performance. The speedup observed with jBridge might have other reasons. More here: https://news.ycombinator.com/item?id=25068976
My experience from actually writing low latency schedulers in user space as well as the publicly available material - like in Ardour - suggests different conclusions from yours.
Keep in mind that a naive benchmark like "cpu usage" is entirely meaningless. What you look at is round trip latency required for a threshold of underruns/missed deadlines. Threading requires additional latency, and process synchronization even more. While I'm sure you report fewer underruns when splitting off into sandboxed plugins I'm suspicious if it's hitting the same performance as doubling or tripling the buffer size in terms of latency in the first place.