To use pci passthrough you have to configure several eareas:

1. Enable iommu, more or less well done via MachineConfig. Not very handy for an virtualization admin but works. An the decision matrix is quite easy, intel or amd… 
2. Configure vfio-pci to allocate a specific pci device
3. In case another kernel driver is already responsible for the device, vfio-pci can not allocate the device. Only option as far as I know is to blacklist the specific kernel driver. To avoid the allocation. 
4. After that, config that a VM’s use a pci device works like charm. Via webui or yaml.

Points 1,2 and 3 need more gui guidance and improvements. Because for an VM administrator in RHEV or VMware it is just a few clicks to select a specific PCI device on passthrough to an virtual machine.

Additional, blacklisting a whole kernel driver means I can not use any other device from the same driver anymore. 

One example: I have 4 NICs (same chipset) first 2 I want to use for OpenShift and second 2 I would like to use pci passthrough to attach the nic’s an virtual machine. Right now that’s not possible.