X79 Express And Another New Processor Interface
It won’t surprise anyone that Sandy Bridge-E necessitates yet another new socket interface.
That’s four introductions in three years, if you’re keeping count. First, there was LGA 1366 for Bloomfield, and later the Gulftown design. Then, there was LGA 1156 for Lynnfield and Clarkdale. After that we saw Sandy Bridge shift to an LGA 1155 interface. And now we have LGA 2011.
Intel’s strategy is pretty clear-cut. It’s trying to carve out one path for desktop PC users and another for the SMB server and workstation folks using 1P and 2P platforms. The company just so happens to also give desktop enthusiasts a taste of that same high-end functionality, basically re-badging Xeons as Core i7s.
AMD, on the other hand, is doing everything it can to keep the desktop and server/workstation platforms separate, which is why we’ve seen a seemingly-slower progression from Socket AM2 to AM2+ to AM3 and now AM3+. Over time, the result has been superior interoperability between platforms. The thing you have to remember, though, is that realizing the potential of each new interface still requires a motherboard upgrade.
Why LGA 2011?
If you bought into LGA 1366, you didn’t step down when LGA 1156 launched, and you probably even held off on Sandy Bridge, knowing full well Sandy Bridge-E was slated for launch this year, too. In that context, a shift to LGA 2011 three years after we were first introduced to LGA 1366 is actually pretty reasonable.
A number of (obvious) changes make the transition necessary. To begin, Sandy Bridge-E does away with the three-component platform that defined Nehalem, X58, and the ICH10. Instead, northbridge functionality migrates into the CPU, leaving just the Sandy Bridge-E processor and X79 platform controller hub. Then you have a four-channel memory controller and 40 lanes of PCI Express, requiring a lot more power and signal pins.
X79 Express: Specifications In Flux
Although we’re sure to see multiple versions of the Patsburg platform controller hub, enthusiasts only have to worry about the one that’ll become X79. The rest will be trimmed and tailored to serve the needs of workstation and server customers.
|Processor Interface||LGA 2011|
|Graphics Configurations||2 x 16 and 1 x 81 x 16 and 3 x 81 x 16, 2 x 8, and 2 x 4|
|Overclocking Support||Processor and Memory Overclocking|
|SATA||2 x SATA 6 Gb/s, 4 x SATA 3 Gb/s (6 x Total)|
|Rapid Storage Technology||RAID 0, 1, 5, 10|
|PCI Express 2.0 (5 GT/s)||8|
As mentioned, the expectation of X79 has changed fairly dramatically since its existence was first acknowledged. It was supposed to have a dedicated four-lane storage-oriented link to the CPU, it was supposed to have a lot more storage connectivity, and it was supposed to be SAS-capable.
It looks like we’ll instead see a PCH that sports the same storage package as P67 and Z68: six total SATA ports, two of which run at 6 Gb/s data rates. SAS connectivity is no longer a feature; that’ll instead need to come from a mezzanine or add-in card. Intel also makes it clear that the x4 uplink between Patsburg and Sandy Bridge-E will not be enabled when the platform launches. This might become a reality at some point down the line. Upon introduction, X79 will attach to Sandy Bridge-E solely through a four-lane DMI link.
Patsburg doesn’t support USB 3.0. And now its storage block is looking a little mainstream, too. Consequently, motherboard manufacturers are going to have to use a lot of third-party controllers to get X79-based platforms feature-heavy enough to succeed beefy Z68 boards. Expect at least a handful of the PCH’s eight PCI Express 2.0 lanes to host extras.
The bottom line is that X79 ends up looking a lot like P67 Express. All of the platform’s differentiated functionality comes from the Sandy Bridge-E processor itself.