What's the diference between these two examples? In example one you have assigned 8 vCPUs to the virtual machine and in the second example only 6 vCPUs. But that has nothing to do with any best practices.
If we say that a pCore to vCPU ratio of 1:3 is okay, that would mean that you can distribute a total of 72 vCPUs to all virtual machines on a physical server with 24 cores. Whether a machine gets 6 or 8 depends on how many VMs are running on the ESXi host.