News & Analysis
Comment
PR14
tjwal
The case against Thunderbolt
Rick Merritt
2/28/2011 11:00 AM EST
SAN JOSE – Intel Corp.'s new high-speed I/O technology, Thunderbolt, is leaving some people feeling burned. Rather than drive a new interface into the market, they want to see the chip giant give its full attention to a successful, existing one—USB.
Thunderbolt brings new costs and complexity to deliver two bi-directional 10 Gbit/second copper links that won't open up any major new applications, they say. USB 3.0 is already available at data rates up to 5 Gbits/s over copper and, like Thunderbolt, can also ride optical links in the future.
Simply put, Thunderbolt "is a mistake," said one big Intel customer.
Their arguments—not generally being aired in the public—are the reasons why some of the biggest PC, display and hard disk drive vendors are not backing Thunderbolt so far.
Thunderbolt will clearly carry a price premium, although Intel won't say how much. The controller, currently made only by Intel, will have a cost roughly in line with today's 10 Gbit/s Ethernet chips. That represents a premium over the cost of a USB 3.0 chip.
In addition, Thunderbolt requires a unique five-wire active cable so far supplied only by Apple with modified mini DisplayPort connectors from an unnamed source. Other costs are hidden in the complexities of mastering a new technology, potentially with new supply chain partners.
Apple is so far the only system maker adopting Thunderbolt. It has a business model based on selling generally upscale products, typically with higher than average PC profit margins. Most PC and display makers primarily compete on cost in higher volume markets with thinner profits.
Companies not yet publically backing Thunderbolt say there are no compelling applications that need more than the 5 Gbit/s links USB 3.0 can offer. Intel managers say Thunderbolt is unique in supporting display resolutions greater than high definition, but that's a very limited niche.
Thunderbolt will let OEMs have one port that can support either a display or high-speed data link, potentially simplifying designs, said Intel. But opponents said systems will still need to support existing interfaces such as USB and making room for one more connector on space-constrained systems such as ultrathin laptops will be difficult.
Intel argues Thunderbolt will let OEMs build systems that put previously internal resources such as fast disk drives or graphics outside the box. However, opponents say such designs are already possible with a cabled PCI Express spec, and so far no one has found them compelling.
In the end, Thunderbolt's detractors want the industry to put all its efforts behind USB. It is a relatively low cost, well understood technology, already shipping billions of ports across all computer, peripherals and consumer devices.
The USB 3.0 version was architected in a way that it can be, like Thunderbolt, extended to optical links and higher speeds. Intel has not backed it strongly enough, delaying plans to support it in its PC chip sets, they say.
Thunderbolt will no doubt get more backing from top tier PC, display and disk drive makers. But it remains to be seen whether it will provide a sustained light or flames out like Firewire.
|
|
|
|
|
LarryM99
2/28/2011 11:22 AM EST
I really don't see Thunderbolt as a replacement for USB. If anything, USB3.0 is ahead of the curve a bit in its established peripheral market. What I do see are new uses for it to which USB does not apply. For example, you could use it for a subsystem interconnect to move expansion cards outside of the PC chassis. Need a couple more PCIe slots? Add an external chassis via a Thunderbolt link.
Larry M.
Sign in to Reply
GREAT-Terry
2/28/2011 11:59 AM EST
Higher speed may finally turn out new applications, especially from some companies like Apple. Without knowing the price tag however really puts some uncertainty on how well the market can accept this new technology. But Intel+Apple seems is a good reason why people can keep watching at.
Sign in to Reply
evangellydonut
2/28/2011 12:17 PM EST
The new SSD external HDs using SandForce 2 controllers are pushing 400MB+/sec speed that USB 3.0 is unable to support (400MB is theoretical after protocol overhead). Will USB 4.0 come out in 2012-13 to support the next generation SSD drives? I'm buying a new Macbook Pro because of the fact it supports Thunderbolt so I can use it with the new SSDs and not have an interface bottleneck.
Sign in to Reply
fdunn
2/28/2011 5:22 PM EST
Then don't buy an external HD without either FibreChannel or 10Gb ethernet fabric.
You starting to feel the pain yet?
"Thunderbolt" (really lame name) is nothing more than 10Gb ethernet with a differing protocol. With that said, 10Gb transceivers (optical in particular) are not yet commodity priced interfaces so it will start off slow and more than likely be replaced at the point of "final take off" by 100Gb ethernet.
Sign in to Reply
elctrnx_lyf
2/28/2011 12:22 PM EST
This situation looks like the big leaders are pushing technology even if it is required or not. But the consumers will decide the future of it.
Sign in to Reply
fdunn
2/28/2011 5:27 PM EST
There will be room for both but as you said the consumers will have the final say.
The motherboards I see and buy have many times more USB ports than FireWire due to the formers high take-up rate and low cost.
On the flip-side (as a hardware guy) FireWire does not use as much CPU time as USB.
IMHO - Thunderbolt (LightPeak is better) will be limited to the Mac domain and there will be some PCIe Add-On cards for PCs.
Sign in to Reply
halarpd
2/28/2011 1:39 PM EST
The controller chip is required on both side, in transmitter - a notebook and in receiver - a peripheral. Also, as written here, TB is expanding PCIe to an external box so if you want to put any device on this extend PCIe bus one would need PCIe to xyz converter silicon - for example PCIe to Sata for sata drive. With today's semiconductor technology all these (TB controller plus xyz peripherals) could be integrated in a low cost solution but it does not exist today. So I would say there were will be players like Apple who will use this technology but any mass adaptation is two to three years away. Oh by the way the DP is only 1.1 not 1.2!!
Sign in to Reply
dang99
2/28/2011 7:01 PM EST
DP 1.1a has been replaced by DP 1.2. Several of the OEMs have announced and will be shipping this month DP 1.2 with the Sandy Bridge platforms.
Sign in to Reply
Sanjib.Acharya
2/28/2011 1:40 PM EST
Not only USB 3.0...after Intel comes out with an optical cable for Thunderbolt in 2011, what is going to happen to "Lightpeak"? Has Intel dropped the idea of "Lightpeak"?
Sign in to Reply
Frank Eory
2/28/2011 2:26 PM EST
Thunderbolt is the new name for Light Peak.
Sign in to Reply
Code Monkey
2/28/2011 4:18 PM EST
10GBPS serdes probably fits better into their process roadmap, where 5GBPS (USB) either leaves too much performance on the table or fizzles out too soon. Apple invented USB, so they surely looked long and hard at the tradeoffs between USB and Thunderbolt.
Sign in to Reply
fdunn
2/28/2011 5:34 PM EST
Apple did not invent USB, Intel did.
Designer Ajay Bhatt, Intel
Designed January 1996
Manufacturer Intel, Compaq, Microsoft, NEC, Digital Equipment Corporation, IBM, Nortel
Superseded Serial port, Parallel port, Game port, Apple Desktop Bus, PS/2 connector
Apple is not even on the Board at the USB consortium: www.usb.org
Sign in to Reply
selinz
3/1/2011 11:04 AM EST
Apple needs something to differentiate themselves because the consumers are beginning to realize that they are other prodcuts offering the same functionality (or more) that are more accessible (price and useage). I just don't think this is much of an attraction.
Sign in to Reply
Jeff Dickey
3/7/2011 5:06 AM EST
I expect the second rev of TB to be compelling, certainly more so than USB 3/4. Apple, for instance, has had a long history of first-gen products that show amazing potential, while leaving it to the second-gen products to start delivering on that. Think Thin Mac/Fat Mac, Apple I/Apple II, iPad/iPad 2.
Apple already have a speed differentiator; in my testing, FireWire 800 delivers over twice the throughput as USB 2. This is true both on Apple gear and on tests I've made using a "mainline" PC desktop with a dodgy Chinese FW800 PCIe card in it.
I'd love to see a real horse race develop in I/O technologies. I think one of the problems with the PC for the last several years has been the pervasive monoculture of USB 2, and training people to not expect more efficient/higher speed interfaces when there was dubious-at-best engineering rationale for not delivering them.
TB has its problems. (One display device? and if I want to change, I have to do *what*?) But so did USB 1.0, back in the day. Intel learned and did better, and I'm sure they will again. If not, Apple will find some other hardware partner who will. They've shown, more than once, great willingness to make major, shift-the-earth changes when progress wasn't as fast as they thought it should have been.
While it may be true, as Yogi said, that "the future ain't what it used to be," there's little reason to doubt that it's going to be very interesting.
Sign in to Reply
Bear1959
3/4/2011 10:12 AM EST
I predict Thunderbolt will end up being Vaporbolt.
Sign in to Reply
StevePxxxxxxxxxxxxxxxxxx
3/4/2011 4:32 PM EST
The merits and utility of the interface don't matter.
It's a positioning game.
Intel chip-sets have an interface that will likely get massive co-op advertising.
Customers will see the port on some products.
Enter "fear uncertainty and doubt".
For $10USD more I get a feature I don't understand but has been getting a lot of advertising.
Many will make the "safe choice".
Sign in to Reply
damngoodengineer
3/7/2011 1:07 PM EST
I don't see the point of such a negative article. You really think we'd be better off without it, waiting for some alternative that we haven't heard of yet? Would you have rather seen Apple come out with USB3 ports? What?
Rather, why don't you bitch about how this came about with such apparent exclusivity.
Sign in to Reply
tjwal
3/7/2011 3:16 PM EST
“640K ought to be enough for anybody.” -Bill Gates (1981)
Sign in to Reply
David Ashton
3/7/2011 3:57 PM EST
And it would have been, for a long time, without Microsoft's code bloat.....
Sign in to Reply
tjwal
3/8/2011 3:51 PM EST
David
Can't argue with you there, but the following quotes would have worked just as well
"I think there is a world market for maybe five computers.
Thomas Watson
President IBM 1943"
"Everything that can be invented has been invented."
-- Charles Duell, Commissioner of US Patent Office, 1899"
Point being that the arguments that there is no current need for thunderbolt and therefore it won't succeed are short sighted.
Its success is likely more dependent on the inertia of the market than on any current need.
Sign in to Reply
PR14
1/10/2012 2:11 AM EST
"Apple is so far the only system maker adopting Thunderbolt"
Sony Z series laptops.
Sign in to Reply