How does VMWare CPU MHz sharing / reservation work on the host CPU

I'm not sure about servers, but the current thinking in embedded devices is not to bother with steps between low-power and flat-out because the extra time involved will eat your power savings, so basically they run low power until they get any real amount of cpu load at which point they flip over to fastest-possible so they can finish the job and get back to idling at low power.

A vCPU can only be mapped to a single physical CPU. You can't take 4 physical CPUs and make a single vCPU that's 4x faster; it's just not how it works.

Hyper-V is limited to assigning 4 vCPUs to a VM (last I checked). If you need significant CPU power, go physical, there's no sense in adding virtualization overhead to something that CPU intensive and parallel in the first place.

Also, as Holocryptic notes, if you assign 4 vCPUs to a VM, that VM can't run until Hyper-V has acquired 4 physical CPU cores to run them on. Depending on your configuration this could be a major stumbling block (ex, if you have a 6-core machine with a bunch of 4 vCPU VMs, only one will ever run at a time, the other two cores will always go essentially unused). According to Jake Oshins this was not true for any version of Hyper-V. He states that Hyper-V does not use gang scheduling for the CPU; as almost every other hypervisor does. Accordingly, if one phystical CPU core is available, Hyper-V may use it to run a multi-CPU VM. (Also mentioned, Hyper-V may not use all the physical cores available at the time because of NUMA partitioning)

Side note: SQL doesn't necessarily use all the cores you can throw at it in the first place. It really depends on what you're using it for and how parallelizable the load is.