Why do multichannel plugins always max out a single core in Nuendo?

The only time I ever have CPU issues is every time I work in Atmos. Im a Dolby Atmos engineer. So that’s a problem.
When I create 7.1.2 reverbs/delays in Nuendo13, I see the resources jump way up as soon as the first 7.1.2 reverb is applied. It doesn’t matter whether its Pro-R2, Blackhole, Seventh Heaven, or multiple instances a mono/stereo/quad/5.1 reverb (Nuendo insert routing); the same thing always happens as soon as I get to a total of (3) 7.1.2 reverb/delays busses:
Nuendo resources say its overloading and the freezing up or inability to play confirms that its true, but Task Manager always says CPU and RAM are below 50%.

I’ve removed all processing from everything else, and it still happens.
When I have just (1) 7.1.2 Reverb, (1) Reverb on an FX channel of 5.1 or less, and (1) 7.1.2 Delay, Nuendo’s resource meter teeters on the edge of an overload, but I can usually squeeze by.
I can then add analog and plugin inserts to the tracks & channels without issue.
HOWEVER, sends become an issue; especially SC sends.
It doesn’t matter whether the song has 6 tracks with my usual 2 plugins per track, or 126 with 5 plugins per tracks.

It seems that either the resource meter is way off and acting up or Nuendo is sending way too much to a single core.

I have the multicore option activated because the PC is running a 16-core CPU. I thought this was the type of thing that option is supposed to prevent.

Just curious, what happens if you turn that option off?
What does the CPU manual state, as far as this type of processing, if the manual states anything at all?

I remember long ago, we had to only allow a maximum of N - 1 cores on our systems when running Pro Tools TDM on the early multi core Macs. Where you had to leave one CPU core free for the operating system, or it would constantly crash. I know, different brand / processor type. But, possibly a similar situation.

I haven’t seen a Nuendo manual since awhile before the option was added, but I recall Steinberg marketing the new feature, at the time, as being the answer for studios that had been asking for it to utilize AMD Threadripper/TRx-Pro CPUs’ high core counts.

The only software that experiences any issues when Im running 2+ immersive reverbs/delays in Nuendo is Nuendo. I can open Bing and search for stuff with 5 tabs open, reprogram the console controls in the remote app, fiddle with Dolby Atmos Renderer Remote, recall analog settings in Neve Recall, check task manager, etc. while Nuendo has a hissy fit about the workload. So, Windows itself has plenty of CPU resources to do whatever it needs.

We’re due for an upgrade, but the old AMD Ryzen9 5950X we’re still running offers a 3.4GHz base clock, 4.9GHz Turbo/5.0GHz XFR), 16cores/32thread; which is still quite capable by today’s standards.
In fact, the thing still keeping us from upgrading is that this appears to be a single-core issue and jumping up to a 32c/64t Threadripper would only boost our base clock to 4.0GHz. Since most of our cores have over 50% of their resources remaining when it happens, doubling the cores doesn’t seem like it would help. There are 64 and even 96 core Threadripper Pro CPUs available, thatd allow me to build a 96c/192t monstrous PC with (4)8TB NVMe SSDs mounted to the motherboard in M2 slots, with 7 PCIe lanes, (2) NICs (for Dante/internet), WiFi7, Bluetooth 5, a few Thunderbolt4/USB4, plenty of USB3 ports, 1200W PSU, (4) display ports and 2TB of DDR5 RAM for about $16k [which makes Apple’s $50k Mac Pro look like a lump of dog excrement]…
but if our base clock needs to be higher because Nuendo doesn’t divy out the CPU load for immersive plugins well enough, we’d be wasting money just to end up with the problem getting worse.

I’m sorry, I meant the CPU manual. Cheers.