mache.parallel.topology.thread_siblings

mache.parallel.topology.thread_siblings(cpu_ids: Iterable[int]) → Dict[int, Tuple[int, ...]][source]

The ids that share a physical core with each of these, itself included.

Parameters:

cpu_ids (iterable of int) – The ids to look up

Returns:

siblings (dict of int to tuple of int) – For each id, every id on the same core in ascending order. An id the kernel says nothing about maps to itself alone.