1

I am looking for a CPU to perform real-time computing on KVM with RTOS clients.

For true real-time with the lowest latency possible, 1 VM need to have 1 physical core (to avoid waiting when both machines want to access the same core).

Almost all Intel CPU claiming 16 cores are in reality split in half performance / efficiency cores.

Does the frequency difference between the cores cause problems when two KVM client talk to each other (one having the performance core and the second the efficiency core) ?

Improve this question
4
  • The differences between the two cores, wouldn't typically be exposed down to an application (there are exceptions like kernel level DRM drivers), threat management would be handled by between the kernel and the processor itself. The concept of efficient and performance cores isn't a new thing. ARM has had something similar for years.
    – Ramhound
    Commented Mar 3, 2022 at 14:58
  • Realtime processing in a VM seems quite an interesting concept alltogether - the Hypervisor adds a layer of unpredictability and latency.
    – Eugen Rieck
    Commented Mar 3, 2022 at 16:52
  • Seems to be a LAB solution which might have severe problems with timing; "real time" simulation might not be possible(!?), unless "TIME" also is virtualized (=is not ACTUAL time) .
    – Hannu
    Commented Mar 3, 2022 at 17:35
  • The objective is to simulate the behaviour of the services hosted in the VMs. The time need to be the same accross all VMs (virtual time) and also with the real world (for HIL). The hypervisor is here to avoid a simulation requiering a fleet of physical computer.
    – user161458
    Commented Mar 4, 2022 at 7:19

0

Reset to default

You must log in to answer this question.

Browse other questions tagged .