Our experience is, that you can’t expect full GigE Performance with GigE Vision Cameras on virtual machines.
If you want to use it for testing purposes you need to reduce the bandwidth of the image stream depending on the camera e.g. via Device Link Throughput Limit.
Using GigE Cameras in a virtual environment is always less reliable than on the host system. So I would not recommend using it for more than testing purposes.
Even if the Hypervisors have different implementations, it is always the case that they need to virtualize the network access somehow which results in a reduced performance.
From my knowledge we did not make tests to compare the performance of different virtual machines. Therefore, i can´t tell you if another Hypervisor performance better.
Even then we recommend optimize the acquisition so that you have the lowest possible CPU load and that there is enough room available for additional processing. This means do not use a virtual machine if performance is important.