How would you do this when running in kubernetes? Afaik it just guarantees that you get scheduled on some cpu for the “right” amount of time. Not on which cpu that is.

I only know that there is one scheduler that gives you dedicated cores if you set the request and limit both to equal multiples of 1000.