@Cisco10956 from reading your OP, in this thread, and your two other recent postings, I suspect some of your confusion might be due to you possibly believing PCs, on Ethernet, work "differently" depending on whether they are connected to a hub or a switch (or perhaps also sharing the same Ethernet physical wire, whether p2p between a pair of PCs, or using 10Base2 or 10Base5).
Logically, L3 (e.g. IP) and L2 (e.g. Ethernet) work the same regardless whether hubs and/or switch and/or etc., are being used.
When one host wishes to communicate with other host, it needs the other host's L3 address (e.g. an IP address). However, logically communication of L3 depends on L2, so the sending host also needs the other host's L2 address (e.g. an Ethernet MAC).
First thing the host will do, is map a L3 address to a L2 address. It does this first by checking if already has a statically defined L3 to L2 address mapping. If not, for efficiency, it keeps a cache of recently learned L3 to L2 address mappings, and checks that. If L3 still isn't known, it attempt to acquire that information by sending out a broadcast (L3 and/or L2) directed to all other hosts requesting the L2 address for a specific L3 address (in IP this is a ARP [address resolution protocol] request). (NB: if the L3 to L2 is provided, by the destination host replying, the information is stored in a cache so this process doesn't need to be repeated for every subsequent transmission to the same destination [for a short time period]).
Only if the sending host "knows" or can resolve L3 to L2, can it generate a L2 frame directed to the destination host.
Notice in the forgoing, shared media (i.e. same physical wire), hubs and/or switches, are not mentioned because the hosts don't care, i.e. it doesn't matter.
Shared media, hubs, and/or switches, do matter, in how frames are transported between hosts, but again, not to hosts (at least logically) themselves.
