Intel recently introduced the Thunderbolt connectivity technology. In a 12-month exclusive deal, Apple is the first company to utilize the technology inside the new generation of MacBook Pro notebooks. which is already available for sale. Both companies proudly touted the benefits and advantages of the technology that combines both PCI Express and DisplayPort in a single interface delivering up to 10 Gigabits per second per channel (1.25GB/s). This means you can have up to 20 Gbps, 2.5 GB/s bidirectional bandwidth at each given moment.
Even though these numbers look much better on paper than for instance, eSATA or USB 3.0, this article offers a bit different "food for thought," mostly based on product roadmaps we?ve discussed with several players in the industry.
Key question about any technology is "Where is the catch?" First of all, the 20Gbps of bidirectionally available bandwidth is shared between the DisplayPort and the PCI Express Interfaces. Presenters claimed that you can count on twice the bandwidth of USB 3.0 (i.e. 5Gbps times two), but that all depends on type of devices connected on sides of the cable. Consider a 1080p display connected via Thunderbolt. Depending on a specific DisplayPort mode, approx. 5Gbps of bandwidth will be allocated for a single display. Should you connect another display in the chain, remaining bandwidth would not be dramatically wider than USB 3.0 anymore. We spoke with an unnamed display manufacturer who told us they could eat up complete bandwidth offered by Thunderbolt with their next-gen displays – and they?re not the only one coming to market with small size 2560×1440 (1440p) and desktop-sized 4K (3840×2160, 2160p) LCD panels.
When you attach two displays or use higher resolutions, the remaining bandwidth for other devices is even lower, leaving only a small spark of the once mighty Thunderbolt. Also it should be noted, that it won’t be able to drive any DisplayPort display. Resolutions above 2560×1600 require more than 10Gbps of bandwidth – or go with 10-bit professional grade displays – that is out of the question for Thunderbolt. In a way, optical Thunderbolt with 100Gbps / 12.5GB/s of bandwidth was a step in the right direction, but Intel postponed the optical route until some time in the future.
My next problem with Thunderbolt is, that it’s yet another proprietary interface. Admittedly for display connectivity – it is backwards compatible with DisplayPort 1.2, though only the Mini version currently used by Apple and AMD. Any other type of device either needs a replacement or an adapter to use it. Thus, even if you have financial resources for new add-on devices or plan to buy new ones anyway, these might not be compatible with your old computer anymore (or computers of friends, relatives, etc.). Thus, you?re going head to head against USB 3.0, which is backwards compatible with USB 2.0 and 1.1, without any doubt world?s most widespread interface (RS-232 i.e. serial port takes second place due to explosion of phones equipped with USB 2.0).
Good use for Thunderbolt are NAS (Network Attached Storage) devices and displays. It would be great to see graphics cards that serve as data transfer hubs, such as Thunderbolt-equipped Radeon HD 5870 Eyefinity Edition, with six mini-DP outputs. Of course, the problem is that Thunderbolt is Intel-only technology for now.
With the arrival of Thunderbolt, present interfaces won’t go away – even Intel acknowledges that. But even with Intel?s gigantic market strength, there is a chicken and egg problem with Thunderbolt, which could go the way of FireWire 800 – only being supported on a rather small amount of premium devices. This aspect makes it perfect for Apple though. Apple always sought after exclusive technologies they can offer their customers. Since Apple uses much of the same hardware as other PC makers nowadays, Thunderbolt is something where they can provide additional value – bringing in additional revenue to the company. It’s probably much the same as it was with FireWire 800. While Apple might stick it in most if not all their computers, other manufacturers might take a more conservative approach and only add it to a small amount of premium devices. At the same time, USB 3.0 will be in most computers shipping in 2012 ? making it ubiquitous. Don?t forget that there are companies that use USB 2.0 and 3.0 interface to drive displays already: DisplayLink for instance.
There is one more gripe Thunderbolt or shall I say Light Peak caused me. While this is not officially confirmed, it’s (indirectly) evident that Intel delayed the introduction of USB 3.0 in order to weaken the competition for Thunderbolt. As a major member of the USB Implementers Forum, and co-founder of the standard – Intel has a lot of influence on its development. While they didn’t cripple the standard, they simply didn’t add support for it in their chipsets and made hell to other companies that wanted to implement USB 3.0 inside theirs. The first Intel chipset adding USB 3.0 support will only launch in 2012. As a consequence, third parties like NEC rub their hands in glee as most mainboards and notebooks sold in 2010 sport their companion chip to add USB 3.0 support. If you?re wondering about AMD and why AMD did not go ahead with adding USB 3.0 to their chipset way before Intel, the problem that other chipset makers had (according to our highly ranked sources at AMD and nVidia) was that Intel did not disclose key PHY information needed to implement USB 3.0 inside the chipsets. If you recall, this caused quite a war of words between Intel, nVidia and AMD/ATI back in 2007.
For conclusion of this "other angle" article, we want to state that we are not making Thunderbolt worse than it is. One can look at it as a way to put a video output connection to additional use. The ability to daisy-chain devices certainly sounds nice, though all of these devices need to support Thunderbolt. But other than that it doesn’t bring a lot of new to the table, that’s not already possible with ExpressCard or USB 2.0 and 3.0, including the already mentioned daisy-chaining displays. It provides more bandwidth in some cases, but the question will be whether there is actual mainstream market demand. DisplayPort and HDMI are the future, as VGA and LVDS are being phased out. If anything, it’s a creative way to let customers pay again for a new interface.
Can Thunderbolt gain wide adoption or end up going the FireWire 800 route… that remains to be seen.