In fixed partitioning, what is the occupancy rule for each memory partition?

In contiguous memory allocation the operating system divides main memory into regions called partitions, and the partition is the unit of allocation: memory is handed out one whole partition at a time. Fixed (static) partitioning settles the number of partitions and their boundaries at system-generation time, before any process arrives, and those boundaries are never moved, split or merged while the system runs.

Applying that to the assignment step: because a boundary cannot move, a partition is either wholly free or wholly occupied. The loader places an admitted process into one partition, and that process owns the entire partition until it terminates or is swapped out.

1. The partition table fixes k partitions of sizes S1, S2, …, Sk, decided before any process is admitted.

2. A process of size p is admitted into a partition whose size Si is at least p, and it takes that partition as a unit.

3. The unused Si − p bytes stay trapped inside that partition and cannot be given to a second process. That trapped space is internal fragmentation, the characteristic waste of the scheme.

4. At most k processes can be resident at the same time, so the partition count bounds the degree of multiprogramming.

Cross-check against the other mechanisms named in this item; each governs a different decision:

A placement heuristic such as best-fit may be layered on top of fixed partitioning to decide which partition to use, but it does not change how many processes a partition can accommodate. Therefore, the occupancy rule in fixed partitioning is one process per partition.