When you use a PCI Express 2.0 slot and a PCI Express 2.0 video card you double the available bandwidth compared to PCI Express 1.0. But that doesn't mean that the performance will double or that the communications between the video card and the motherboard will be at double of the speed.
What happens is that current video cards do not use the full bandwidth available on PCI Express 1.0. For example, from the 2.5 GB/s available, video cards probably won't reach half of this. So when you increase the available bandwidth to 5 GB/s nothing will happen, because the bus wasn't causing a bottleneck in the first place. The performance will remain the same.
The same thing will happen on SLI and CrossFire configurations using external bridges. The communication between the two video cards is done thru this bridge, i.e. the cards don't use the PCI Express bus to exchange data between them. So the same idea applies here.
In my opinion performance difference will be noticed only in two situations:
1. With video cards under SLI or CrossFire that don't use a bridge but the PCI Express bus to make the communication between themselves (this option is available only on low-end models). This is also questionable, as low-end video cards use far less the available bandwidth, so maybe when they are in SLI/CrossFire with PCI Express 1.0 they aren't reaching the maximum available bandwidth anyway.
2. With future very high-end video cards. That is why PCI Express 2.0 is being launched. Remember that when PCI Express was launched the first video cards didn't provide any performance gain over cards based on AGP 8x, but as soon as new GPU generations were released the difference became obvious.