PCIe is everywhere for everything, this board makes a lot of sense, especially if it had a HyperTransport slot for a stick of GDDR5 for that IGP, or just to act as a general purpose dual-ported "L4" type of cache.
ASUS is selling PCIe (x4?) cards with two USB 3.0 and two SATA6G on them for $66. Nothing slower can handle the combined data speed of those ports, which is (12+9.6) over 20gbps theoretically.
Some SSD company with Steve Wozniak involved is selling dual-PCIe x16 SSDs that of course plug directly into such motherboards as this. They claim a throughput per unit so high that it saturates both of the slots. They brag about this as part of a one terabyte/second transfer solution they have dreamed up.
Networking is moving to dual/teamed gigabit LAN and it makes a lot of sense to put 2 to four gigabit LAN connectors on the same card as the SATA6G, USB3 or SSD storage, so you can support iSCSI RAIDs over your multi-gigabit LAN, in other words, a storage area network (SAN). If you don't you will need two PCIe x16 slots just for your 100Gb ethernet connection, which is being standardized already. Makes a lot more sense to stick the drive and ethernet controllers all one one card and let it behave like hardware RAID - can you imagine ordinary users setting up a SAN or teamed heterogeneous connections any other way?
If you want to run six monitors you need only four slots with ATI 5970/80/90. Two gamers with their own UIs running on the same box (not an unreasonable configuration for some) might want that. Most however would want to run only three, which you can do with two slots. Saving four for everything else.
Extra GPUs (and ATI's equivalent to Hybrid SLI is called ATI Hybrid Graphics) are useful in general now that OpenCL is supported everywhere. The real reason to use it however is not the extra IGP GPU you get to keep using, but to run (one or two) monitors off the IGP ports on the motherboard and leave the extra GPUs off when not in use - saving 30 to 90 watts.
So here's a use for all six PCIe slots:
Two slots for a PCIe SSD like Woz is selling.
Two slots for EyeFinity (three monitor) graphics, six GPUs total (three per eye, two per monitor)
Two slots for a four-port 10GLAN, one-port 100GLAN, dual USB 3.0, dual SATA6G internal.SAN+RAID card that supports any combination of internal and external drives using iSCSI in any array configuration, and supports any other drives that the OS supports likewise (including the PCIe SSD that can pretend to be dozens of drives if it wants to be for RAID purposes). All of which is about balances that two-slot SSD.
Four DDR3 slots means in practice 16GB, which is just about right to buffer transfers between a couple of hundred gig of SSD and a couple of thousand gig of spinning platters, and one or two gig of GDDR5 (ATI uses this on its high end cards).
AMD's chipset (785 and 800 series) only does about 4000MHz now and at 16-bit that's not fast enough to keep up with those slots loaded. That is, assuming it can actually run all six slots at the full x16. If so the real bottleneck on this configuration is going to be the lack of 32-bit transfer on the HyperTransport bus that AMD uses to connect the processor (the standard bus that replaced the proprietary FSB BS) and a lack of L3 cache on some AMD processors. Configurations like this really need two CPU sockets.
Why not ? Everything else on the box is going to be teamed/paired/mirrored.
Wikipedia has the specs on the 800 series and all other AMD chipsets, at least as far as announced... you should look there before speculating on what *might* be the case about its PCIe combinations. I really doubt more than two slots are there for cooling only.
As for intel it's toast in the high end desktop market: Apple is rejecting its in-core GPU, the FTC is after it for monopolies, it had to cancel both QuickPath Interconnect and Larrabee (graphics card), it won't put USB3.0 support into its chipsets for a year, it thinks PCIe 3.0 (only twice the speed of 2.0) is competitive with HyperTransport (it isn't, especially not with vendors like Broadcom supporting it fully), and it doesn't have any way to exploit the trend to GPGPU (general purpose GPU computing using things like OpenCL) while AMD+ATI is perfectly positioned for this. The danger is that AMD+ATI is going to end up as a near monopoly on high end desktops where all the innovation gets stress tested, leaving intel playing around in niche markets like optical processing and massively parallel supercomputers and NAS appliances/routers (where ASUS will eat its lunch). nVidia also is falling far behind ATI on DX11, on 3-monitor, and on speed.
In 2010 and 2011 you're going to need very good reasons not to buy AMD+ATI, given boards like this one.
