How does PCIe 2.0 relate to PCIe 1.x

nope. The overall bandwidth available to the PCI express slots combined = doubles with 2.0. PCIe1.1 could do 2.5 Gbit/sec, PCI 2.0 doubled that to 5G., but an x16 slot doesn't get any more bandwidth than it used to.

http://en.wikipedia.org/wiki/PCI_Express

Really till the video card manufacturers make a card to fully take advantage of the PCIe 2.0 standard it doens't really make any difference. Testing right here on Tom's Hardware shows that all the current cards that say they support the new 2.0 standard really don't make use of it.

By example the Asus P5K/Wifi has the following pci express slots
x16
x16(x4 signal)
x1(qty 2)
By default if you use the x16(x4 signal) slot for a second video card, it disables both x1 slots because there simply are no lanes available for them in the PCIe 1.1 standard. You can only use up to 20 lanes for the cards.

PCIe 2.0 with the same slot configuration can share 32 Lanes among the cards, meaning you can do a crossfire configuration and still have lanes left over for sound, network, raid and tv tuner cards. Its enough for most of us, but PCIe 3.0 will bring about enough bandwidth for us to have all PCIe slots, and put the PCI bus to rest alongside ISA, PATA and the floppy drive.

SLI PCIe 1.1 boards are a bit different because the 2 video cards run in x8 mode leaving those extra x1 slots open for use in many cases.

Please correct me if I'm wrong. The combined lane limitation of 20 is the limitation of the board and not PCIe 1.1. Even if this board changed all 4 PCIe slots to 2.0, the limit can still be 20 if ASUS chooses to.

And I'm not quite clear about the second x16 slot (1st is blue, 2nd is black, 3rd and 4th are white). The black one can either operate at x16, x4, or x1. If I can only use up to 20 lanes, the combinations are:

Blue @ 16 + Black @4
Blue @ 16 + Black @ 1 + White @ 1 + White @ 1
Blue @ 8 + Black @ 4 + White @ 1 + White @ 1
Blue @ 4 + Black @ 16
Blue @ 4 + Black @ 4 + White @ 1 + White @ 1
Blue @ 4 + Black @ 4 + White @ 1 + White @ 1
and etc.

You said that if Black was to be used as second video card, it disables both x1 slots. In this case, I'm assuming this is the first case of the above list. Blue is used as the first video card, so Blue takes up x16, which means Black can only use x4. The second video card must be weak enough to not request bandwidth greater than x4. Am I right?

Thanks.

the primary graphics does not vary with the p35 chipset, it is stuck at x16 as per that generation crossfire definition. Nvidia chipsets can run at x8, and therefore the rules are different than Intel and AMD's chipsets.

The hypertransport revision works in tandem with how many lanes are available in your chipset. Hypertransport 3.0 sped up the link speed, allowing additional bandwidth between the processor, northbridge, and southbridge = more pcie lanes. Video cards running in a x4 slot benefit from the bridges for additional communication over the top of the cards.

PCIe 1.1 chipset like the p35 has x16 lanes going through the northbridge and x6 additional lanes going through the southbridge.
http://www.virtual-hideout.net/reviews/ASUS_P5K3-Deluxe...

PCIe 2.0 chipset like the x38 chipset has x32 lanes for video going through the northbridge and still x6 going through the southbridge.
http://www.dinoxpc.com/Tests/articoli/articolo/schedema...

Its a lot easier to describe what you can do if you had specific boards in mind that can be addressed.

I didn't know that x16 lane could be fixed on certain chipset. I thought it could always be used as x1, x4, x8, or x16.

Would you comment on my previous question though? On the ASUS board you used as the example, how would the lanes be used if two video cards were used? Would it be x16 and x4?

I had MSI P7N SLI Platinum in mind. According to its official spec, this board does:
Support Dual PCI Express 16X Gen2 interface with 1x16 & 1x8 operation.

It can only do x16 at most per slot. If PCIe 2.0 doubles the bandwidth, which means it can do up to x32 per slot, what is the point of having Gen 2 interface if it only uses up to x16? PCIe 1.x x16 is same as PCIe 2.0 x16.

I feel like I am misunderstanding something fundamental.