SAN JOSE, Calif. – Apple announced its latest MacBook Pro notebooks will build in Thunderbolt, the new high speed systems interface from Intel formerly known as Light Peak. The move will send shock waves through a PC community gearing up to support USB 3.0.
Intel said it developed the technology but collaborated with Apple on bringing it first to the MacBook Pro. The interface sports two bi-directional 10 Gbit/second channels and has a flexible range of uses.
The interconnect supports two protocols, PCI Express and DisplayPort, Intel said. Apple suggested it can also handle Ethernet, FireWire or USB traffic for external peripherals such as RAID arrays and works with adapters for HDMI, DVI and VGA.
The interconnect supports copper or optical cables, daisy-chaining of up to seven devices and has its own native protocol drivers, Intel said. It can also support power over the cable.
The copper versions support three meter lengths at 10W and optical links support tens of meters, Intel said. Thunderbolt supports 8 nanosecond synchronization.
"Thunderbolt technology is expected to be widely adopted as a new standard for high performance I/O," Apple said in a press release. “Thunderbolt is a revolutionary new I/O technology that can support every important I/O standard which is ideal for the new MacBook Pro,” added Philip Schiller, Apple’s senior vice president of worldwide product marketing.
Apple has a mixed history of embracing new system interconnects.
The Cupertino company was among the first to champion FireWire which never gained traction beyond a niche of professional A/V systems and high end disk drives, eclipsed by USB. However, Apple was also among the early adopters of Wi-Fi now standard across all notebooks.
Intel first announced Light Peak in September 2009 as an optical interconnect positioned as a successor to USB 3.0. Last September it said it had accelerated its work and would deliver a controller chip by the end of 2010.
Then Intel went quiet about its plans. Reports emerged it had revised its work to focus on a copper-based implementation. The adoption by Apple, which apparently sought an exclusive deal to be the first OEM to use the I/O, likely required Intel to keep its plans quiet.
The new interconnect could leapfrog work on USB 3.0 which aims to deliver about 5 Gbits/s max and has no capabilities for flexibly supporting multiple protocols.
USB is assured a long life based on its broad adoption across computer and consumer peripherals. However Thunderbolt could eclipse USB as the new high-end interconnect in a similar fashion to what observers once expected would happen with FireWire.
After lengthy delays, mainly attributed to foot dragging by Intel, both Intel and AMD are now preparing PC chip sets that integrate support for USB 3.0 and are expected to be in production in early 2012.
For that scenario to play out, Thunderbolt will have to establish a broad ecosystem of chips and supporting systems and peripherals. Intel is expected to detail its Thunderbolt technology and plans later today.
Intel said companies planning to support Thunderbolt include Aja, Apogee, Avid, Blackmagic, LaCie, Promise and Western Digital. Intel aims to enable the interface for use on other computers, displays, storage devices, audio/video devices, cameras, docking stations and more, the company said.
The Apple MacBooks have also adopted Intel's latest dual- and quad-core SandyBridge processors. They also sport Apple's FaceTime integrated camera and video conferencing capabilities.
The new MacBook line includes a 13-inch model starting at $1,199 which uses Intel Core i5 and Core i7 dual-core processors up to 2.7 GHz and Intel HD Graphics 3000. Models with 15- and 17-inch displays use quad-core Core i7 processors up to 2.3 GHz and AMD Radeon HD graphics processors at prices up to $2,499.
The MacBooks use Apple's Mac OS X Snow Leopard operating system.
Don't overlook the obvious. Thunderbolt can readily do USB, Ethernet, and several other protocols. Apple has always been big on square white box adapters to plug into their more obscure i/o interfaces, for example iPod-to-USB, DisplayPort-to-HDMI, etc. Thunderbolt is all about the PHY. For Intel, it is getting the market primed for the new interface by introducing it using copper. That just means they don't yet have their on-die optical components up to production quality. Once that happens, optical controllers will be cheap and lower power, and they can release optical Thunderbolt into an existing market. That will scale Thunderbolt to at least 100 Gbps, probably to 1 Tbps, and without greatly increasing power consumption / thermal envelope and without decreasing cable distances. Copper interfaces will not be able to compete. Apple will sell a Thunderbolt-to-multiple-USB3 white box adapter to accomodate legacy devices.
Hey Rick ... like the beard! Anyway ... yes, Thunderbolt is cutting USB 3 off at the pass. It's *so* much better a match to the internal connections *within* the system (PCI-E). Systems already have to bridge from PCI-E to USB, and the process is really ugly, resulting in lots of software overhead and truly ugly latency numbers. One of the reasons USB 2, which is nominally 20% faster than FW400, is actually 50% slower in the real world is that poor match between the USB architecture and the memory-bus model of PCI-E, while Firewire *is* a memory bus, so bridging the two is much easier with *much* less software overhead.
Sigh. No ... it's not 10GbE ... it's a fairly generic serdes, but one that is closer to that used in PCI-E. And yes, they could have used an Ethernet infrastructure, but that's not the point ... the point was to have an I/O interconnect that is *software* compatible with the highest speed interface *inside* the system, and that's PCI-E. (The video interface is a great addon, and allowed them to repurpose an existing high speed serial link (Display Port) and a nice small connector (mini-DP).
So, yes, there is little that is *really* new here, it's a wonderful way to bring really high bandwidth *and* low latency I/O to the outside of the box without a big change in software architecture.
And I *really* know what I'm talking about here, since I was Apple's tech lead for Firewire back in the days, *and* I did some USB 2 work after that, *and* I'm now working with the most leading-edge Ethernet stuff around (which is still the best way to do high speed *networking*, but not necessarily I/O).
And I think "Thunderbolt" is a great name. USB, blech ... eSATA, give me a break ... FibreChannel, *so* geeky ... "LightPeak", better but still blah ... remember that "Firewire" was a name that people still remember, while "iLink" (the Sony term) has been thankfully forgotten ... more details (if you care) at http://johasteener.com/firewire_FAQ.html#Why_all_these_names
Apple will do just fine. I think Steve & co learned their lesson with Firewire (technical superiority can't compete with lower cost and wide adoption) ... similar to the Sony Betamax lesson. Note that this is NOT an Apple initiative ... it's an Intel initiative with *input* from Apple. Apple could have just as easily adopted USB 3, another Intel initiative ... but they have way too much sense to buy into that particular boondoggle.
Actually, I know a *lot* of moms and pops that have no idea what *any* of the connectors are for, except for power. Their kids know about USB as the "iPod" connector. Aside from that, connectors and cables are geek territory, I'm afraid ... and among the semi-geeky types, the USB connector is hated because they can't figure out which way to plug it in. Of course, the truely geek know that the USB symbol has to be "up" (or whichever direction "up" might be for vertical orientations, or the bizarre way they are mounted on my NEC display). Anyway, USB has run out of steam, and USB 3 is really kind of an embarrassment (something that I tried very hard to reorient into a more reasonable direction, but the Intel USB guys didn't want to listen). Thunderbolt, on the other hand, is really a great combination of Intel technology and technomarketing muscle and Apple's sense of style and what a *consumer* might like.
quite true. It will take some effort to convince customers, which have been listening to Apple dearly for sometime. But it will be the industry players, which will be the make or break of any technology. Widespread support is required for any technology to be successful.
I think this a huge win for Intel and Thunderbolt. Thunderbolt is a technology with a huge upside especially when you look at the optical implementation. While USB 3.0 is competitive at this point, Thunderbolt is at the beginning of the growth curve and we are squeezing the last bits of performance out of USB.
For Intel to have this in the market and out of R&D, starts to validate it and give it legs. If we remember how long it took for USB to gain penetration, Thunderbolt needed a win to start the process of acceptance.
Without backward compatibility with existing USB 2.0 devices, no hope. Mistake by Apple. With USB 3 ports you can totally remove USB 2.0 ports. With Thunderbolt, you still need to keep them.
My guess is, Intel Bribed Apple big time to adopt this non standard tech. Something like 25$ + Free chip/mac for every thunderbolt deployment.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.