Each vCPU would be scheduled to process on a CPU core. CPU scheduling can be a complex topic, but essentially ESXi would have to find 4 free cores at the same time to run a CPU cycle for your vCPU VM. With your 2 x dual core server the VM would be running on all 4 cores. On the server with 8 cores, at most the VM could max out 4 out of 8 cores if you were running the VM at 100%. Which CPU cores would depend on the ESXi scheduler.
