Locking a Qube to specific processor cores

[nephtonlabs]

Oddly specific question for a weird use case but here goes:

Is it possible to lock the virtual processor of a Qube to a specific physical core of the host system?

From what I see it looks like the underlying Xen hypervisor supports it (1), but I wanted to check whether anyone has tried it in Qubes, and had any luck with it.

Long story short the 7950X3D and 9950X3D processors have 3D V-Cache on one of two of the CCD chiplettes and I was toying with the idea of whether you could spin up a VM, lock it to just the cores on that chiplette, and dedicate a virtual psudo-R5 or R7 X3D processor to that Qube.

I guess another use case might be banishing a low-priority Qube to an Intel e-core, though I suspect that might have issues.

(1) Virtualization: configuration options and settings | Virtualization Guide | SLES 15 SP6

Feel free to tell me why I’m stupid. This was just spitballing a weird idea for gaming optimization.

[Privacc]

I think you are talking about cpu core pinning / CPU affinity. Search the forum for those terms, you’ll find a lot of information.

I’m interested how vcache will be handled. I have a pc with a vcach chip, but I don’t run QubesOS on it. If you do get it working, let us know what your experience with it is.

[corporateblush]

Hybrid CPU: benchmarking performance when pinning to specific cores

Related to hybrid CPUs with Performance/Efficient cores and this topic CPU Pinning Alder Lake explaining how to assign CPUs to a Qubes OS VM, I made a benchmark showing the performance difference between 4 performance / 4 efficient / 4 cores with no special pinning in Qubes OS I added a Ryzen 5 5600X desktop CPU and a Laptop with an i5-7300U (Lenovo Thinkpad T470) to the benchmarks. The T470 CPU is only 2 cores with hyperthreading, so it has 4 threads at best, sharing with the single benchmark …

… and from there following links posted, might help.

Also

Qubes Cpu Pinning: automatically pin CPUs for VMs

I was referring to this comment by @renehoj, wherein you could help make SMT (moderately) more secure by setting sched-gran=core, but it was not possible with asymmetric CPUs. I am not fully sure about that, althought the Xen wiki page about CPU pinning does not mention that it should only work with credit(1). Later in the day I can do some testing based on the amount of fan noise/temperature rise when I perform soft pinning.