Hello Mark,
a point to point subinterface has the property that only one DLCI is associated with it and allows to accomodate a point to point logical connection between two routers acting as DTE. Multiple point to point subinterfaces can be configured under a single interface each with its own IP subnet ( typically a /30 is used to save in address space).
A point to multipoint subinterface can have multiple DLCIs reaching multiple remote routers. All routers connected to the p2mp subinterface have an IP address in the same IP subnet.
Actually the physical interface is multipoint in nature. The Inverse ARP protocol allows to determine what IP address can be reached at the remote end of each DLCI.
The use of subinterfaces has influence on routing protocols for different aspects.
EIGRP and RIPv2 being distance vector protocol may suffer of split horizon issues.
OSPF require special treatment of point to multipoint topologies as any-to-any connectivity is not possible so placement of Designated Router on the hub router is important.
OSPF has multiple options of network type that have different usage.
In real world the use of point to point subinterfaces is more common.
The interaction between Inverse ARP, routing protocols in p2mp topologies is often present in certification questions and labs.
Edit:
the frame-relay map commands that you have on configuration are important for the broadcast keyword that allows to support broadcast and multicast that includes the discovery of routing protocols that use multicast based hello packets.
In order to support routing protocols in p2mp context you need to put the broadcast keyword in all mappings.
Inverse ARP supports only the resolution of the unicast WAN IP address of each DTE router.
Hope to help
Giuseppe
