Why Thunderbolt Is Dead In the Water

adeelarshad82 writes "In the same way that Apple championed FireWire for the replacement of parallel SCSI, Thunderbolt is meant as the next big thing in video and audio peripheral interfaces. Plus, it's Apple's move to beat USB 3.0. However, Thunderbolt is off to a slow start, for a number of reasons — from cost to the technology's features in comparison to USB 3.0 — which is why it may be dead in the water."

Comments

Re:Really?

[dgatwood]

Actually, that's not true. Thunderbolt provides a significant win over USB 3 in nearly every way. The author just doesn't get it.

First, Thunderbolt is based on PCIe for transport. That means that it's a very lightweight protocol, unlike USB, which is very heavyweight. For things like audio interfaces, USB 3 is dead in the water because it offers no advantages over USB 2 (because throughput doesn't matter past a certain point). Thunderbolt, by contrast, should offer a significant advantage in latency over FireWire (and a huge advantage over USB 2), while requiring less CPU overhead than USB or FireWire.

Second, it's entirely unclear to me why anyone supports USB 3 at all. For hard drives and similar, USB 3 offers no advantages over eSATA. For almost all other devices, USB 3 offers no advantages over USB 2. So ignoring portable devices that only have room for one port, USB 3 is a solution in search of a problem.

Third, the author doesn't know what he's saying about copper being "crippled". It's not crippled at all. Thunderbolt is intended to eventually be supplemented with new cables that have an optical PHY (transceiver) inside the cable instead of on the logic board. Such a design provides exactly the same advantages as LP (distance), but without all the problems that optical interconnects inherently suffer. To describe thunderbolt as "crippled" because it uses wires is to fail to understand the technology at all. It's exactly as fast as Light Peak was originally intended to be for its initial rollout.

Fourth, using LP in a USB connector turns out to be a bad idea in general. USB is a great interconnect for low bandwidth devices. It's not so great for talking to displays. With desktops tending to go under the desk, and with more and more people using laptops with external displays at home, there's good reason for wanting all of your external devices to be plugged into your display. Sharing a single data connection for your display cable and your peripherals is a tremendous win—so much so that support for transport of USB data was actually built into the original DisplayPort specification. Thus, Thunderbolt shouldn't be thought of necessarily as a replacement for USB, but rather as a replacement for other display technologies. With Thunderbolt, you could trivially build a monitor that provides full-performance, low-latency FireWire, USB, and eSATA connectors on top of your desk. Try that with USB 3.0.

Finally, the cost of Thunderbolt hardware is probably greatly exaggerated. Sure, it probably does cost $90 to add TB into a motherboard design right now, but that's because A. it isn't integrated into the motherboard chipsets yet (wait for Ivy Bridge), and B. it likely requires a significant board redesign to free up enough PCIe lanes to support the metric crapton of bandwidth involved.

Thunderbolt will become a lot more interesting when Intel starts integrating it into their chipsets in Ivy Bridge. Until then, it's really not feasible to most folks to start using it yet. Thus, it's not at all a surprise that adoption has been slow. Right now, it's basically at the developer preview stage, with AFAIK exactly one working motherboard implementation (Apple's).... The author should at least wait until Ivy Bridge before making predictions about the technology....