why would it be a bottleneck? Right now we're looking at maybe 10x-12x Pci_e 1.1 is used, not even the entire x16, you won't lose any performance by not going to pci_e 2.0