Back in the day.... when you setup VMs to have, for example, 4vCPUs, the host server needed to have 4vCPUs actually AVAILABLE in order for the hypervisor to schedule a CPU cycle for the VM. Which meant, assigning 4vCPUs on a VM -may- yield worse performance than assigning 2vCPUs, if the host server is a lot of VMs an limited cored. (in other words, don't over-commit vCPUs if not necessary!)
In 6.0+, has this scheduling process changed? Is there more flexibility to allow for better performance on VMs that require a high number of vCPUs, or has this remained the same as it always was?