i cant see any reason why not , after all its just a generic PCIe x4 and these have been available since and for PCIe 1.0a Endpoint Controllers.
even a simple dual core ARM cortex, never mind a Quad core Freescale A9 on a cheap PCM along side these Thunderbolt Controllers plus whatever else you care to connect up from the SOC should work fine.
although you dont need to max out the data throughput from any single device/protocol OC as its multi protocol capable as well as up to 7 devices daisy chain-able as it stands today from what we are told.
it really is frustrating that to date no x86 OEM has bothered to order and integrate these Thunderbolt Controllers and ports, after all, they have generally known about this coming for a long time now, and yet not even a basic generic PCIe x4 Thunderbolt card is available
why didn't Intel make a PCIe x4 Thunderbolt card batch or two and have them available and on sale as a market tester from day one i wonder
OOOPS, I meant 10Gb/sec in the second para above (SFP+ Copper Interconnects)!
That aside, Thunderbolt has to improve on the mechanical dimension challenges of the USB-3.0 while not compromising on the data fidelity. The Intel website for Thunderbolt mentions that the interconnects are daisy-chainable; one would have to use retimers for this.
One of the biggest things going for Thunderbolt would be the compatibility with Displayport which opens up a huge market in signage.
@rick.merritt: Who are the Silicon players for the Thunderbolt interconnects? I would be curious to know... in contrast, there are quite a few Si vendors for USB 3.0.
@david.may: you can get 5m reach at Gb/sec in copper TwinAx without pre-emphasis/equalizer chips at 30AWG and longer with the chips (15m possible). This configuration includes one lane for Tx, one for Rx, differential signaling.
With 10Gig Base-T, the patch cords can go much longer; with AV bridging IEEE is re-promoting, it may not be too long before we see consumer appliances offering 10Gig RJ45 ports.
Yes, I'm interested as to how competitors will respond. I wonder if they've had anything in the works that could be similar to the Intel / Apple technology. For example, AMD, which has traditionally made competitive chips could partner with some of the OEM's not yet on this technology to move forward. This article was excellent asking all of the relevant questions about this exciting new technology. Once specs are released, I'm interested to hear about security measures to protect devices with this technology.
I would have to agree that getting to the article information was quite time consuming. When the next page does come up and is not chock a block full that is also disappointing. Good details, but presentation/format was not helpful..
" thin notebooks or clients can use Thunderbolt to link to high-end drives, displays and other external devices instead of building them all into one box. Essentially, it provides a PCI Express link outside the box "
So is this the missing link for cloud computing where apple didnt show up? Can we assume apple is having plans to build its cloud Os based on T.bolt interconnection?
It looks like Intel learned a few things about marketing serial technologies. Firewire was never incorporated into it's chipsets so it never became a commodity technology. The licensing pool for Firewire was not nearly as effective as USB; it had to be called 1394 for a long time because Apple wouldn't release the name "Firewire" for many years.
The implication from the statements that Thunderbolt implements PCIe as a protocol, indicates that it's a substantially different serial transport than Firewire or USB. Software will easily be an order of magnitude more complex.
Microsoft has made great strides with driver support in general but supporting bus drivers is well beyond the demonstrated capabilities of OEMs or ODMs. Linux will be challenged by this without strong direction by Intel. Of course, Apple will do what ever it wants and it will work.
Best on the record for USB/Firewire adoption, it's easily 3 years before Thunderbolt really starts to impact the market and 5+ before it dominates.
Blog Doing Math in FPGAs Tom Burke 23 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...