How does CPU affinity affect Linux Real time guarantees?

Question

I am envisioning a linux architecture like this:

                    --------------------
                   | LEGACY APPLICATION |
----------------------------------------
| AMAZING APPS     | LINUX KVM PLATFORM |
-----------------------------------------
| LINUX KERNEL WITH RT_PREEMPT PATCH    |
-----------------------------------------

Now this envisioned architecture is to achieve "Wrapping in of new functionality" around the old application.

For example:

1. I would like to take a new linux hardware, like super fast I9 12 Core system.
2. Install a Linux based virtual machine on it which emulates the hardware which the legacy application sat on.

3. Then run the legacy application on that virtualized platform (So the inputs/outputs would still be the same).

4. Run my new "Amazing App" system directly on the linux kernel.

Now:

1. If I set CPU affinity of "linux kvm platform / Legacy Application" to 6 out of 12 cores. (taskset #1)
2. Set CPU affinity of "Amazing Apps" to other 6 cores. (taskset #2)

My question is:

How will this affect the real time guarantees of "Legacy Application"? I will install RT_PREEMPT patch, but not sure if this assumes availability of all the CPU cores?