Today is the big “coming out” day for Thunderbolt (nee Light Peak), courtesy of Intel and Apple’s new lineup of MacBook Pros (one of which is sitting on my desk). Next week (March 2, to be exact) is the introduction of another “Magical and Revolutionary” Apple product, the iPad 2. Inevitably, pundits are putting 2 and 2 together and deducing that the future iPad will include this new I/O port. But this makes little sense. The iPad 2 won’t include Thunderbolt.
What Does Thunderbolt Need?
Thunderbolt is a very high performance technology that works at a very low level. It’s nothing like USB or BlueTooth, which include everything from physical-layer signals (electrical and radio frequency, respectively) to application-layer interfaces. Thunderbolt is strictly a transport for higher-level protocols. Although it includes some network-type functionality (it can be switched and even routed a bit), it’s not the sort of protocol that would communicate directly with peripherals.
Curious about Thunderbolt technology? Read my detailed description!
Apart from DisplayPort-compatible displays, Thunderbolt peripherals will need a built-in PCI-to-something chip to do any real work. They’ll include an Intel Thunderbolt “decoder” chip to de-multiplex and repeat Thunderbolt signals as well as some other kind of controller. For example, LaCie’s cool new Little Big Disk must include a PCI-to-SATA RAID controller chip that presents its Intel SSDs to the Mac it’s connected to. As one Intel spokesman put it today, Thunderbolt devices will look like they are inside the computer.
At least for the time being, any computer with a Thunderbolt port will need to use the same Intel “Thunderbolt Controller” chip found in my MacBook Pro. And this controller chip requires two very high-end-computer things:
- Four PCI Express v2 lanes
- A DisplayPort video signal
This is why, as Intel clarified at today’s press conference, we will not be seeing Thunderbolt add-in cards any time soon. Even a native PCI Express bus doesn’t have DisplayPort, and most computers don’t have four spare PCI Express lanes lying around anyway!
Thunderbolt is probably something of a power hog as well. According to spec, it can provide up to 10 Watts to connected peripherals (enough to quick-charge an iPad, natch). But pushing 10 gigabits per second over a copper cable takes quite a bit of power as well – just ask any Ethernet switch maker! I’ll guess a Thunderbolt port running at full speed will draw another 10 Watts or so from the host just for data transfer.
What Needs Thunderbolt?
We should also consider why a Thunderbolt would be useful for a given computer. The most obvious use of this new high-performance I/O port is to connect to large storage devices or high-resolution video displays. Indeed, these are the self-same use cases illustrated in Apple and Intel literature.
Another potential use case is a port replicator or docking station scenario, where Thunderbolt is used to extend the system bus. Imagine a laptop computer accessing an ExpressCard slot, some USB ports, and an Ethernet drop over a tiny cable shared with the display and you get the idea.
Again, we should stress that any Thunderbolt peripheral will likely match at least a few of the following conditions:
- High-performance in terms of throughput (like a disk drive) or bandwidth (a monitor)
- Large or expensive enough to justify the cost of multiple ports and controller chips
- Complicated enough to require something more than a USB or FireWire connection
This is why we are unlikely to see Thunderbolt media card readers (except perhaps in the case of pro video formats like P2) or flash drives (except perhaps in the case of big, fast SSDs). USB is just fine for mundane tasks like reading SD cards or transferring files, and consumers will balk at the tens of additional dollars all that Thunderbolt hardware will cost.
The iPad Doesn’t Need Thunderbolt
This is why the iPad 2 will be neither a Thunderbolt host or peripheral: It wouldn’t make any sense.
I can’t think of any reasonable use case for connecting an iPad to a big, fast, expensive external device like LaCie’s Little Big Disk or even a DisplayPort monitor. An SD card slot would be nice, but connecting that with Thunderbolt would be like driving a Hummer to the grocery store! (Wait a second…)
Then there is the fact that the iPad lacks the kind of internal hardware needed to be a Thunderbolt host. I doubt it has any internal PCI Express lanes, let alone four to spare. And it is unlikely that it uses DisplayPort internally, though it would be kind of handy to have some kind of digital video output better than VGA.
The idea of the iPad being a Thunderbolt peripheral is a little less far-fetched. After all, the dozens of folks foolish enough to pony up for a brand new MacBook Pro today (hey now!) might want something to connect to. And syncing over Thunderbolt would be awfully quick! Plus, Thunderbolt could charge the iPad as quickly as the 2.1 Amp power brick that comes with it!
But the iPad exists in a competitive marketplace, and Thunderbolt can’t be high on the list of priorities. It needs a dual-core CPU, more RAM, and two cameras to keep up with the competition. High-performance I/O isn’t really a “must-have” feature.
The iPad 2 will not be a Thunderbolt “host”. There’s no way Apple could shoehorn in the whole Intel controller at this point, and there’s no need for it either. Wait for the iPad 4. Or the iPad 8.
Although I’m confident the iPad 2 won’t be a Thunderbolt peripheral either, I can imagine one really good reason for Apple to do something that foolish: Pride. Syncing a tablet at 10 Gb/s is just the sort of “because we can” nonsense we sometimes get from Cupertino, and it would be a nice excuse to talk about this new technology in public.
Regardless, I’m not holding my breath. I expect a Thunderbolt Cinema Display, MacBook Air, and Mac Pro to come along years before a Thunderbolt iPad!