Original link in German
Translated to english with Google. 
 

 
This better be a bombshell to those who've disagreed with me for the past year...I'm not about to claim vindication however, consumers may never get this board -- note italtics; there is no reason we can't get a single socket version.  My thinking is simply rooted in that

1. there is a large gap between the LGA1155 (16 lane dual-channel DDR3) and LGA2011 (40-lane quad-channel DDR3) that it would make sense eventually we get a single LGA1356 (24-lane triple-channel DDR3) setup.  If it comes, it will be after LGA2011, to funnel as many sales into the larger socket first, to get it paid off. 
2. Maybe I'm the only one who noticed that the consumer nomenclature of Sandy Bridge DT ended at 600 with i7 2600?  In the previous, there was a 700 and 800 in LGA1156, then i7 900 in LGA1366.  If LGA2011 gets the i7 2900 series to signify absolute high-end, what about the 2700 and 2800?  But if we only see the three upcoming SBE CPUs as i7 2900, that IMO confirms it. 
3. Consider the chipset rumors thus far, the highest-end mainstream chipset is called Z77, yet the LGA2011 chipset for desktop market is called X79 -- what about 78?  maybe Z77 will have an Ivy Bridge DT put into "2700" for a 22nm quad-core or something, X78 may take an i7 2800 for LGA1356, while all Core i7 2900's will be LGA2011 X79's only.

Mark my words: I strongly suspect that we'll see later on an "X78" single LGA1356 setup.  

Okay, I was doing some thinking on how Intel might make a hypothetical "X78" appealing; and it may have something to do with Ivy Bridge.
 
In terms of debut schedule, mainstrem Sandy Bridge came first as desktop and laptop, and this past month a Xeon flavor came (E3-1200's).  We are expected to see X79 this fall/winter, I think November which was when Nehalem's X58 first came.  Those CPUs are rumored to be a limited-OC 3.6GHz quad and two 6-cores are 3.2GHz and 3.3GHz with more OC flexibility, respectively.  Then the Ivy Bridge mainstream parts come to succeed the Sandy Bridge couterpart parts.  I think the highest end chipset , Z77, will have to 22nm CPUs for it: an i7 2700 to succeed i7 2600.  But I don't know if it will be a K model.  My thinking is that 2500K/2600K will be milked and most of the Ivy Bridges will be in lower markets.
 
I think right after that X78 comes with only Ivy Bridge parts to set the appeal apart from X79's 32nm parts with more PCIe lanes and their premiums.  I'm thinking at least one limited-OC 22nm quad like 2700 model but with 24-lanes of PCIe 3.0, instead of 16.  There has to be an unlocked LGA1356 part, not as expensive as the X's, maybe priced like the cheapest 6-core LGA2011, but I'll bet it will be a quad-core.
 
If it were a 6-core, it would cannablize the X79 6-core especially after people realize they won't need 40 lanes of PCIe.  I think the 6-cores will stay in X79 initially, then by next winter 2012, Ivy Bridge will bring 8-cores to X79 for those who didn't switch to X78.  After that 6-cores can some to X78, though I'm sure Intel would milk their X79 as much as possible.
 
So tentatively, it looks like LGA1155 will take up to quads with 16 lanes, LGA2011 will have 6-8 cores with 40 lanes, and LGA1356 may come to fill the gap with bclk OC quads but 22nm initially just to get them sold or appealing.
 
Intel has done this before, if anyone remembers when Core 2 debuted in summer 2006, at the time they were loosing market share to AMD and needed more than one SKU to take that back, so they had five from 1,86GHz dual-core to the 2.93GHz unlocked dual at $999.  But, their first quad stood alone for months at $999 (QX6700) and the first 45nm part also stood alone for months at $999 (QX9650).  So when Intel has you, they have you.
 
With Nehalem, the socket change and lack of a competitor (nForce variant) meant they needed to sell the socket, hence three Bloomfield flavors.  Of course they were too close together, which was why other models were brought to space them out.  When Gulftown appeared, again it stood alone as the only $999 option for X58.  But then rumors about AMD getting Bulldozer remotely on time meant that Intel had to debut mainstream first instead of traditional high-end first since AMD wasn't competing with X58.
 
Now with X79 upcoming in a new socket, Intel will once again introduce multiple flavors to sell the socket.  I think they positioned them right this time, especially reducing the amount of overclocking at the cheapest $300 level to entice the appeal of the cheaper 6-core at $500.  It may seem like unfair pricing, but IMO this is another clue that something else will fill this gap.
 
The 3.6GHz quad already threatens i7 2600 by price and specs, even though they are both limited in overclocking, maybe the same 4x max adjustment, but the LGA2011 part has the 40 lanes.  An i7 2700 part would displace some of the fire stolen by the LGA2011 3.6GHz quad, while an i7 2800 in LGA1356 could be the gap filler between the two LGA2011 parts if it allowed flexible overclocking.  I figure it could be priced between, in the $400 range and spec'd at 3.2GHz.
 
While X78 could threaten X79 for a few months, then the 8-core models for X79 come and problem is solved.
 
If an single-LGA1356 based "X78" is to appear, it will be between Z77 (april) and X79's shrinks (winter 2012), so I'd say about June 2012.
 
Definitely, the dual 1356 boards will come earlier, like March.  All dual 1366 boards show EOL (product discontinuance) on Intel's since in Q2 2012, three years after they were debuted. 
 
Anyway, just my $0.02 speculation.

"You can tell by the slots"? Is that like, "I can tell by the pixels"?

Naming Scheme
 
Retail
SB->IB
LGA1155 2600->3700
LGA2011 2800->3800
LGA2011(Dual CPU) 2900->3900
 
My prediction

is it me or does anyone else find the idea of a dual mismatch CPU kind of stupid

What do you guys think of a X78 chipset using single LGA1356 debuting mid-2012?  It couldn't be another LGA2011 which would canniblize X79, while a Z78 would canniblize Z77 in LGA1155.
 
I understand the lack of popularity since everyone is focused on LGA2011, but it flies in the face of the idea that the socket doesn't even exist, which was more or less assumed just because rumors haven't been bringing it up.  As if those who leak rumors aren't choosing the leak whatever they want...
 

SirWaWa
only one full PCI-E slot?
weak
yep, server board

What's funny is all the slots are accounted for, add them up and it is still less than the total of 24 lanes per CPU -- plus they are PCIe 3.0 slots with IIRC twice the bandwith.  So three x8's means three 2.0 x16's, but of course some modification is required to gte as many graphics cards in there.
 

big_willie
is it me or does anyone else find the idea of a dual mismatch CPU kind of stupid

How so, or are you talking about that one other Asus experiment?  Theses sockets use the same ones, flipped 180-deg.
 
BTW, is that Alicia Keys in your sig?

seronx
Naming Scheme
Retail
SB->IB
LGA1155 2600->3700
LGA2011 2800->3800
LGA2011(Dual CPU) 2900->3900
My prediction

Core i9?  Just because it is getting shunk with more cores doesn't require a brandname change, but we'll see I guess.  Tentatively, I would have figured Haswell/Rockwell would get the 3000's and a new set of sockets.

DraginElite

looks like the smurf version of the SR-2 ...

not bad though seeing more of the Intel Line up

 

not to sure if that would server only market item, where are the pciX slots, most server use that type of slot for sas cards and what not. looks more like a proof of concept type product.

I thought that the PCIe 3.X was a function of the CPU with SB-E and IB. The lanes are a simple hardwired path to the Socket IINM.
 
If the above is the case, then no current socket 1366 cpu's could get PCIe 3.0 without a change to the Bridge setup of SB/IB. I'm not sure that the 1366 socket has room for moving the PCIe 3.x on chip. In which case this is simply an overpriced (possibly) X58 with usb 3.0.
 
Sean