Equal share means that everyone shares above their allocated CPU resources (spare cycles). But I would ask specifically on a case by case basis, because sometimes a proportional/weighted share is actually taking place - varies from provider to provider.
These comments are directed toward answering the OP questions. As the OP outlined in his first example where (Mhz / VM) = CPU per VM. This is the essence of the equal CPU share model.
I would offer the following clarifications. In practice what this means is that the host does not apply any type of usage policies as to how individual VMs may access the available cpu cycles. That is, they are up for grabs by the individual VM owners.
What this means in real life is your experience will vary between two extremes.
All VMs will use little CPU because none are running CPU intensive applications. Use is primarily interactive offers the best performance for everyone.
Mixed Environment: some percentage of the VMs run CPU intensive applications. This offers good performance for the heavy CPU users because they may only be sharing the CPU cycles with >50% of the users which means that the heavy CPU users can expect relatively high-to-consistent performance. e.g. 3000Mhz machine with 4 heavy CPU users( ~750Mhz heavy per user) obviously, a little bit less when others are online and more during windows when any of the 4 needs less cycles.
Finally, the worse case scenario where all or more that 50% of the users are heavy CPU users AND they are using cycles constantly 24/7 resulting in most users receiving the minimum allocations at all times. Therefore, in the last case your actual number of CPU cycles will depend on two factors.
1). The number VM on the server however, keep in mind that this isn't always a good indicator of use because of #2
2). The actual load the VMs are placing on the host server.