I'm unsure how this will pan out. However, Thunderbolt is architected in a manner to extend PCIe protocol outside the box (high bandwidth, low (8ns!) latency). This is a significant differentiator for many new usage models, whereby a laptop (an increasingly large share of the computing market) could provide the brains for a stationary expansion chassis at each work area.
To the comparison with 10 GbE - no, it is not identical. Without going into the protocol improvements in relation to Ethernet, 10GbE is 10 gigabits/s, but _half_duplex_. Thunderbolt is full duplex, with 10 Gb in each direction simultaneously.
USB3? Fixed master/slave relationship makes it unsuitable for many scenarios that Thunderbolt would be a good fit.
Lastly, Thunderbolt is now what Intel is calling this technology - it is not an Apple-owned name. 'Light Peak' is now officially a 'project code name'.
No, another point - the transport protocol is as important as the fact it runs at 10Gb (x2). This data rate will doubtless increase as time goes by. BTW, what's a 10GbE link cost today? Is it anywhere near the price of a 13" MacBook Pro with (admittedly only one end of) a Thunderbolt link ($1199)?
OK, one final point - the so-called Mac Tax seems to have vanished. When I was last looking at laptops, any first-tier i7, 17", decent graphics, comparably equipped laptop from HP, Lenovo, etc. cost as much or more as the MacBook Pro I ended up with (my first-ever purchase of an Apple as a main machine, BTW). No, Apple doesn't make a $300 laptop. But apples to apples ;) they're competitive in several market segments.
Some interesting comments here, especially from the detractors. However, how many display devices will use a USB3 interface to the PC? And can you run HDMI over USB3?
I think there's plenty of room for both Thunderbolt and USB, especially since USB already has such a wide acceptance (but then so did the floppy disc...).
If they are so innovative then why are they essentially copying 10Gb ethernet?
I think they are still using a copper interconnect because an optical interconnect at that bandwidth is expensive and power hungry, especially if it has to drive that bandwith any significant distance.
There is nothing new about this, except the protocol, otherwise everything was already here and available.
Just put another NIC in the device and have it run 10Gb ethernet, you can do the same thing as with (oooohhh) THUNDERBOLT!
I'm sorry but who ever thought up that name at Apple should be canned.
Thunderbolt amounts to 10Gb ethernet.
Big whoop, we can do it across 10G ethernet as well as this interface.
It's not optical like Apple and Intel discussed and that was the major criteria that Jobs wanted to differentiate it from USB3.
USB 3 is already shipping. I'm with Rick on this one. It may be faster after it ramps up but so was firewire and it lost to USB. I'm not saying it's not around but not nearly as prevelant.
You ask any mom or pop that uses a computer what USB is and they will know. On the other hand unless they got in early and needed it for the "old" Videocams (you know the ones that used tape ;) they won't know what you are talking about.
Thunderbolt costs significantly more than USB3. Most technology companies are not interested in having Intel dictate what they do, specifically in the cell phone industry and SoC business. Apple has a remarkable track record over the past 5 years with consumers. Beyond that, the support list is not strong since it lacks major OEMs. It sounds like Thunderbolt will be successful depending on what Apple has coming out in the next year or two. USB3 will have a large following. Many OEM products in the pipeline have support for both USB3 and USB2. Overall, I think both USB3 and Thunderbolt are strong contenders.
PCIE may be a power hog, but just ask video card manufacturers if 10Gb/s is enough for them. Those companies currently make 16 lane PCIE Gen2 cards. That's 5GT/s * 16 lanes * 4/5 (due to 8B/10B coding) = 64Gb/s. And PCIe Gen3 is coming this year (that will be 8GT/s * 16 * 128/130 = 126Gb/s)! So I agree that Light Peak should be a good high speed external interface, but PCIE is not threatened yet inside the box.
Don't forget why FireWire never went much of anywhere. Steve Jobs insisted on excessive royalties that prompted the development of USB 2.0 to take over the roles FireWire would otherwise have owned. Apple doesn't have that sort of control over Light Peak aka Thunderbolt.
The real question becomes how best to allocate transistor real estate. Would it be more cost effective to have USB 3.0 implemented as a Thunderbolt to USB hub device on the motherboard, along with software to do much of the heavy lifting? I can foresee a system in a few years that appears to have a full complement of ports but they all connects to Thunderbolt controllers under the hood.
It is primarily a high-bandwidth external link, fast enough for high-resolution monitors and for box to box communications, when you need to send fast data more than 1 meter away. USB will still handle audio, non-critical memory storage, etc., and PCIe will handle internal links in the box. I remember that USB used to be called Unused Serial Bus for years until it took off. This could be a replacement for HDMI.
Blog Doing Math in FPGAs Tom Burke 24 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 ...