How come the official spec is for 1 display + 6 devices and the macbook pros and iMacs (maybe mac mini too?) can do 6 devices + 2 displays?
It's probably because it's dual channel. 6+1 per channel. The second channel is probably all DisplayPort.
No, it is 6 TB devices total on all the chains (implicitly the primary host comptuer makes the 7th). You can have a DP dangling on the end.
TB is an aggregator. You're not gonig to have lots of aggregators on a chain. One, they tend to be more expensive so not going to buy many. Two, if are expensive high bandwidth devices you'll run out of bandwidth.
How is the 2 channel version going to work? No bi directional or no displayport?
All Thunderbolt ports and cables can carry 2 x 10 Gbps bidirectional channels. Each direction in each channel can carry data and/or display. There are two versions of the host controller, one that supports connections for 1 port (2 channels) and another that supports connections for 2 ports (4 channels).
Intel has been consistent with saying that the maximum number of devices than can be connected in a chain is 7. This makes sense if you think about it, the controller on the host PC being ID 0, and the attached devices IDs 1-7 (all very base 2). In addition to Thunderbolt devices, you are also allowed to attach native DisplayPort 1.1a displays, but they have to be the last device in the chain, because there is no way to attach another device to them to extend the chain. This would allow for one attached DP 1.1a display device where the host PC has one TB port, and two DP 1.1a display devices where the PC has two TB ports. Additionally, with the imminent arrival of the Apple Thunderbolt Display, you will be able to daisy chain up to two displays off of a single port.
This is where things become a bit sketchy with the available info, though. Intel's initial literature clearly states you can add "up to a total of 7 devices, 1 or 2 of which can be high-resolution DisplayPort v1.1a displays (depending on the controller configuration in the host PC)" This would seem to imply that regardless of the number of TB ports available on the host PC, each TB controller can only address 7 devices in total. What is not so clear from this statement is whether the limitation on attaching native DP 1.1a devices arises from the physical number of ports on the host PC, or instead from the number of DisplayPort connections the host PC can provide to the TB controller. In reality, it could potentially be limited by both. This also fails to address whether you could successfully connect 4 displays by using at least 2 ATDs to a host PC with a single 2 port Thunderbolt controller.
To further confuse things, this most recent slide states: "Daisy chain topologies - 6 Thunderbolt devices and 1 native DisplayPort display." If I had to guess, I'd reckon that this was merely provided as an example of a daisy chain that would work on any Thunderbolt enabled PC. I still believe that you could have 7 Thunderbolt devices (if you could find and afford them) connected in one chain without a DP display in the mix.
It would appear that there is not a lot of distinction at this point between the silicon used by host PCs and that being used by devices. (i.e. the chips are the same, man.)
So will the Thunderbird chip be native with Ivy Bridge?
No. Quoted from Anandtech: "While USB 3.0 will finally be integrated into the chipset, Thunderbolt will not. Intel clarified that the interface will be featured on some 2012 platforms but it wouldn't be on all and it won't be integrated into the chipset."
Agree 100%. This opens the possibility of a user accidentally using a Mini DisplayPort cable instead of a Thunderbolt cable to try to connect a hard drive or whatever. Sure, there are markings on the cable, but markings often wear off. It's always best to change the physical connection (unless you are able to use the exact same physical wires, like from USB 1.1 to USB 2.0).
I pondered this as well, but then I came to the realization that the number of people who actually have a mini-DP male to mini-DP male cable kicking around in a draw somewhere represent such a small minority that this will probably never be a significant problem. I'm guessing Apple came to the same conclusion.
If the only reason to produce fiber-optic cables is to get 10 meter lengths, with no speed increase, but a huge cost increase, I don't see the point. How many people are there out there really that are saying, "you know, I'd pay almost anything to have this thunderbolt enabled hard disk 10 meters away from me!".
Maybe there's demand I just don't see.
Up the speeds to 10x their current value, and you'll see lots of interested buyers, however. IMHO They should wait until the fiber-optic cables can get faster.
For people who would pay all outdoors for a Thunderbolt to 10 GbE or Fibre Channel adapter, or A/V pros with obscene budgets, there could be a good reason to put some distance between you and your Thunderbolt attached device. (I'm thinking research labs, data centers, movie sets, recording studios...) TB is not really about HDDs so much as it is about fringe applications.
Thunderbolt is currently limited by the back end, not the cable, and I don't think Intel wants to be too vocal about this point right now. (i.e. come out and say that the real bottleneck is DMI 2.0, the general lack of sufficient PCIe 2.0 lanes with current mainstream chipsets, or any shortcomings of their integrated HD graphics.)