The PCI-E 3.0 Debacle Explained.
Moving on, let's talk about PCI-E 3.0 and how manufacturers can claim as such. On a typical Sandy Bridge motherboard the 16 PCI-E lanes from the CPU are routed to 2 slots for multi-GPU capability. The first 16x slot on most boards (the one that can actually run at 16x with only a single card) has the first 8x of the slot directly wired to the CPU's PCI-E controller. The second set of 8x pins in that first slot can be switched between the second half of the first slot or the first half of the second slot. The catch is that the second 8X from the PCI-E controller has to go through PCI-E switches to change between the slots. Now the issue lies within those switches, as those switches are built for a certain amount of bandwidth, typically on Sandy Bridge boards they are built to handle PCI-E 2.0. Now if you see 4 PCI-E switches like above, each of these can handle a total of 2 X PCI-E lanes (2 lanes x 4 switches=8 switchable lanes). In the case of this board, the PCI-E 2.0 switches are swapped out for ones with PCI-E 3.0 specs. Here is how you tell if it’s a PCI-E 3.0 or PCI-E 2.0 switch when dealing with Pericom PCI-E switches like the ones above:
PCI-E 2.0: PI3PCIE2
PCI-E 3.0: PI3PCIE3
The last digit above is the first digit on the second line of those switches. The number 3 denotes PCI-E 3.0. Simple stuff right? Well not so much it seems, as there is a lot of controversy over whether or not existing boards with PCI-E 2.0 switches can support PCI-E 3.0 at all, and well while ASUS and GIGABYTE claim they can have some sort of PCI-E 3.0 capability (the first 8x that is directly wired to the first slot), companies like MSI and ASRock are putting out materials to disprove them. There is no doubt that if the lanes are directly wired (not through switches) to the slot then they can support PCI-E 3.0. The issue comes when half the first slot runs at different speed than the other half. There is also some consideration that the PCI-E 3.0 spec asks for different resistors and capacitors, but to the lowest tolerances required, which both ASUS and GIGABYTE most likely use. We can only wait and see whether or not those boards will support PCI-E 3.0, but we have to wait for Ivy Bridge, and that is even if Ivy Bridge has PCI-E 3.0, as the plans still aren't finalized for that CPU. We do however believe that you, the reader, should be able to make an educated decision on the topic and that is why we are giving you this little briefing on the argument. Consider that after GIGABYTE made its claim, MSI and ASRock came out with counter-claims, then GIGABYTE came out with a counter-counter-claim, and then ASUS came out claiming what GIGABYTE had originally claimed. It might not be over, so grab some popcorn.