Is DMVPN Phase 3 functionality hindered if the routes don't point to the hub?

I understand how the redirect from the hub triggers the spoke to do an NHRP resolution request for the other spoke. I also understand the scalability it brings to the control plane if we do this. However what I am trying to understand is do the resolution requests still get CEF switched and sent to other spokes without having this setup? For example assume we have Phase 3 setup with EIGRP but we don't modify the next-hop. Routing table aside, does this increase the load on the hub or spokes? I ask because in production I see topologies like this all the time.