USB 3.0 for the masses – Dispelling the myths.

There has been a lot of unanswered speculation recently regarding USB 3.0 and Intel’s involvement; I thought it was about time to set the record straight. By the way: USB 3.0 will be a new wired USB standard – operating at faster speeds than previous USB generations.

Rumour 1: Intel is creating the USB 3.0 specification.

Much of the incorrect speculation in the press so far has centred on what the USB 3.0 spec is, and who is creating it.

There are two separate standards being developed, USB 3.0 and Intel’s Host Controller spec in support of the USB 3.0 standard. The recent press articles do not properly distinguish these separate specification development efforts.

USB 3.0 – The spec:

USB 3.0 is not an Intel specification; it is being developed by the USB 3.0 Promoter Group (HP, Intel, MSFT, NEC, NXP, and TI). The USB 3.0 Promoters issued a call for contributors in November 2007 and since then the USB 3.0 Promoter Group has been joined by over 180 USB 3.0 Contributor companies (Including other chipset makers such as AMD and Nvidia) who are helping to finalize the USB 3.0 specification. This spec is expected to be made publicly available by the USB 3.0 Promoter Group along with an adopter agreement early in the second half of 2008. (Very soon)

Intel’s host controller spec – wholly different:

In parallel, but separately; Intel is investing heavily (think gazillions of dollars and bazillions of engineering man hours) in resources to create an Intel host controllers spec in order to speed time to market of the USB 3.0 technology. Think of the host controller spec as a ‘Dummies Guide’ to building a USB 3.0 compatible piece of silicon; it is NOT the USB 3.0 specification itself. The industry is keen to get this guide as it will allow them to build USB 3.0 compatible circuits without repeating the massive investment undertaken by Intel.

After all, the sooner USB 3.0 hits the market, the sooner all you readers will be flooding your devices and hard drives with insanely large files requiring masses amounts of computational resources, improving your lives, and making you pleased that you bought a quad-core processor.

Intel plans to make this spec available early in second half of 2008 with a no-royalty licensing obligation (Basically: free, gratis, unpaid, zero dollars, free of charge, at no cost, on the house). This isn’t only because we are just nice guys, but it is also because Intel has set the bar for technology leadership and industry stewardship. It is Intel’s stewardship that has lead to USB being the most successful interface in the history of computing. + We at Intel love it when available processor performance is used to the max.

This is the same successful process that was followed for host controller specifications that supported the USB 2.0 and Wireless USB specifications.

Rumour 2: Intel is holding back the specification, and not sharing with the industry.

No Intel isn’t holding back the specification, the whole point of Intel investing heavily (gazillions of dollars and bazillions of man hours) into creating this ‘Dummies Guide’ is to enable the industry to start building USB 3.0 into their silicon as soon as possible, so why would Intel purposefully delay? One danger however of distributing an unfinished spec is the risk of incompatible hardware down the line, leading to a right mess. As an Intel specification Intel has the responsibility to insure that specifications we deliver to the industry are fully developed and mature enough for others to use. The Intel host controller spec is expected to be unveiled to the industry as soon as possible, in the second half of the year.

The impatience of our fellow chipset-makers (as described in the press) to leverage Intel’s investment and begin to design great USB 3.0 supporting devices of their own is however very encouraging and should aid a fast USB 3.0 adoption ramp.

Could anyone invest the necessary time and money to create such a host controller specification? Of course they could – but with Intel’s Industry stewardship in sharing our spec the industry doesn’t have to make the additional time and resource commitment.

Rumour 3: I heard that USB 3.0 borrows technology heavily from the PCI Special Interests group. Is this true?

No, not true. The USB 3.0 specification has not borrowed heavily from the PCI-SIG. As an industry leader, Intel routinely provides technologies and intellectual assets to the industry through participation in standards development organizations, special interest groups and other forums. For example, Intel has provided significant technical input on the PCI Express architecture developed in the PCI-SIG. Similarly, Intel has also provided significant technical input on the USB specifications developed by the various USB Promoter groups over the last decade, including USB 3.0. (Hint: read between the lines, we contributed to both).

33 Responses to USB 3.0 for the masses – Dispelling the myths.

  1. Jen-Hsun Huang says:

    Come on! The douchey-ness here is off the scale:
    “…investing heavily (gazillions of dollars and bazillions of man hours)”
    Obviously someone at Intel who runs the P&L has the data to tell exactly how much time and money has been expended on this, and it is probably a very small percentage of the much, much larger costs they are responsible for. You don’t have to tell us what those costs actually are, but please don’t pretend that it is some incomprehensibly large number! It just makes you look like a douche.

  2. Rick James says:

    So, this is pretty much a way to avoid the horrible disaster that was 802.11n? I don’t know people are so p.o. about this. Isn’t this how USB 2.0 was rolled out? That went pretty smoothly.

  3. um fırat says:

    the previous comment seems intuitively true. c’mon, is the USB technology much more complex and expensive than CPU’s of the next generation? gazillion and bazillions sounds like godzilla on a roller coaster

  4. Igor says:

    Nick has exaggerated when he said a lot of time and money, but USB host controller specification surely is complex, more complex than the USB protocol specification which is complex in itself.
    I personally do not understand what’s the rush? It is not like there are USB 3.0 gadgets available for sale. I also don’t see any Intel boards with USB 3.0 yet. As far as I remember USB 2.0 was launched simultaneously by all major chipset players. Fact that only AMD and NVIDIA are bitching about it while smaller and thus more affected vendors like VIA and SIS are silent clearly shows that someone is crying wolf.

  5. Rich Arzoomanian says:

    I’m glad someone from Intel admits USB is a piece of junk, although he does it in a backhanded way. I always tell people to stick to the PS/2 ports for the keyboard and mouse, because they have low overhead compared to the monstrosity called USB. USB has always been Intel’s way of using up more CPU cycles, and therefore needing a faster processor. It’s a business decision, but it’s a horrible technology that is poor at everything. eSATA or FireWire is much better for hard disks, or anything else like that, the PS/2 ports are better for keyboards and mice. USB is just a lazy person’s way of interfacing something poorly, and a sloppy technology that has too much overhead.
    Not that I dislike Intel. I just don’t like USB. Intel has come up with some great products (286, 486, Pentium Pro/II/III and now the Pentium III+/Core 2).
    I have to say though, whoever thought of the name Core for the processor should be sterilized. It’s like creating a new car and calling it “engine”, or a new air conditioner and calling it “compressor”. Some of the code names like “Katmai” would have been better and more interesting to consumers, especially in Alaska.

  6. Justin Murray says:

    Well, after reading this and other articles and forums about USB 3.0, I’ll just wait for 3.1 to come out……..

  7. Julian says:

    “After all, the sooner USB 3.0 hits the market, the sooner all you readers will be flooding your devices and hard drives with insanely large files requiring masses amounts of computational resources, improving your lives, and making you pleased that you bought a quad-core processor. ”
    so USB3 wastes CPU cycles and I need to buy a quad core CPU to use it… great engineering

  8. Roy Mustang says:

    Good blog post. Informational as well as an informal tone. You may or may not be BSing us with the company line but you have done it in a way that makes you believable.

  9. Ginie says:

    Good info but what speeds/Specs we looking at. If u have a look at the speed jump from USB1 at 12Mbps, at which speed is only good for mice and keyboards; so the official USB body upgraded the specs to USB 2.0, adding Hi-Speed USB mode operating at 480Mbp. Will we see a huge jump or not

  10. IntelNick says:

    Just a clarification:
    - I think USB is brilliant.
    - It is not USB 3.0 which will take up CPU cycles, but rather the manipulation of super large files (think HD video content) that will be transferred onto the computer via USB 3.0.

  11. Nicholas Orr says:

    I didn’t even know USB 3.0 was being developed :/
    I also don’t get why there is complaining from over vendors. Surely if they are apart of the same USB Promoter group then they have the same info that Intel has and could go about making their own Host Controller – so why complain and just go make your own… Oh that’s right you don’t want to invest in that… haha something for free is best complained at rather then doing it yourself, right :)
    Anyway as with all technology these days it’s all old news once it’s released. The consumer just goes about their day and all they know is that if they plug this “thing” into that other “thing”, “stuff” should happen…
    btw Nick I love the read – it’s great to see this sort of stuff on Major Company Official sites, bring it all back to reality, humans are in charge of these companies not robots :)

  12. Oliver says:

    So Intel will develop the HD-USB3 host controller, and AMD, Nvidia etc. will develop BluUSB3? Sounds like a recipe for customer satisfaction – not.

  13. knc says:

    Everything is USB today and motherboards come with up to 12 usb ports built in, that’s pretty good but I still think of USB as only good for things that are slow or temporarily connected.
    eSata is much better for external hard drives if you are going that route.

  14. battery says:

    [...]USB is just a lazy person’s way of interfacing something poorly, and a sloppy technology that has too much overhead.[...]

  15. Mohannad Asayed says:

    We are wait these Greater Technology
    and forever from
    INTEL … the head of all Companies

  16. Mike says:

    “so USB3 wastes CPU cycles and I need to buy a quad core CPU to use it… great engineering”
    No, but dealing with the HD images and videos you’ll be capturing and storing requires lots of CPU cycles.

  17. Nicole says:

    Wow ~ With all the emerging technologies that are available, i.e. FireWire, GigE, CameraLink, PoCL, etc. isn’t it safe to say that the new USB 3.0 probably won’t be adopted by everybody anyway? At least on an industrial level?

  18. elvis tong says:

    early in the second half of 2008? your mean in the second half of 2008 we can see the product of USB 3.0 .I hope so .
    come on intel ,wish you can bring more and more good things to the world

  19. Mike says:

    Intel should not bend to the likes of nVidia, AMD, and VIA. No free ride for any of you. If you don’t have the R&D dollars to participate with your own funds then get the hell out.

  20. John Gilmore says:

    Any competent chip vendor could design an interface chip for USB 3.0; the problem is that if every different chip vendor does so, it requires every software company to write separate drivers for each vendor’s interface chips. In USB1.1, there were two different software interfaces and two drivers. This got hashed out for USB2.0 with a single hardware/software interface and a single driver. Clearly, Intel is angling for a single hw/sw interface and a single driver for USB3.0. This goal is good for consumers and good for the industry.
    The catch is that Intel doesn’t have a monopoly on good designers. A cooperative approach that included some of the software vendors (who end up having to live with whatever Intel designed, whether it’s a monstrosity or a panacea) and some of the other chip vendors would be more likely to result in a high quality, long lasting design.
    That cooperation would involve releasing draft specs so that affected parties (driver authors, other chip vendors) could read them and comment. And considering the use of overall design approaches that were “not invented here” at Intel.
    Instead, Intel seems to intend to drop this as a fait accompli — take it or leave it, warts and all. This lack of cooperation seems designed to give Intel’s chip designers a year of private access in which to build their own chips, while others will have less time, and will reach the market later.
    Reading between the lines, that seems more like what AMD and nVidia are complaining about.

  21. Martin says:

    USB 2.0 was essentially USB 1.0 because 1.0 was too slow for anything except a mouse and keyboard. Look at all the things that we now use that have USB: Flash drives, cameras, hard drives, printers, and test equipment (I had to get that one in).
    USB 2.0 was a step up in complexity over USB 1.1, but it created many devices today that make our lives easier. USB 3.0 at around 5 Mbps will bring a new generation of devices that we can’t imagine yet.
    Having said that, though, there was no need for USB. Firewire has always been way ahead of USB in high speed PC communications and it’s an IEEE standard (1394). USB is not. But, Firewire was invented by Apple and of course Intel couldn’t handle that. What a waste of engineering resources, developing USB when Firewire was already there. Now, there are so many USB devices out there that Firewire has lost. Too bad.

  22. sixscrews says:

    …super large files (think HD video content) that will be transferred onto the computer via USB 3.0….
    Sure, I would love to transfer super large video files via USB 3.0 but I’m nor sure the producers of the super large video files (read: movie studios) would want me to do that, at least without some crippling protection.

  23. nullone01 says:

    Ah, excuse me for being stupid but, WHY??? This will need to be replaced by something better in the future and it appears they are saying that it will not use the same plug/interface; well that is brilliant! Make a new standard based on an antiquated one that is not backward compatible; if that is true (I have not looked into it; just saw that where I got linked here from…) that will be horrible!
    Now, look at SCSI stuff. Reality is, SATA II /eSATA (II) will do one’s arrays OK for next to zero investment. Now look at where SCSI is today, OH YEAH if you check it will still blow everything off of the map, but how much does a 30GB drive cost a company or an individual these days? How many vendors sell any SCSI equipment or hardware? With hard drive technology of today, we are ready to convert our storage to cheap massive RAM ARRAYS. I am surprised this is not happening more… This all looks like a sham and a shame.
    Don’t get me revved up about ASPI layer drivers for SCSI hardware and Microsoft either. That is a mess that really troubles us with our SCSI hardware…
    I am in-line with most of the comments folks have posted, but without reading them all I just got to ask about this guy’s comment about how they have been at the forefront of technology and specification innovation. Does he mention that they are still on the hook to AMD licensing 64-bit technology for their Intel platforms? I don’t think it has been 7 years yet… They don’t like people to know that their 64-bit instructions were written by AMD, unless I got disinformation from my insiders… Take care all! I pity the fool that has USB 1.0, 1.1, 2.0, 3.0 etc. RAMRODDED into their hardware!

  24. Hi Nullone01,
    It will use a backwards compatible plug/interface, just like USB2. No need to panic. :-) You will be able to plug in your USB 2 appliances.
    Nick

  25. -hh says:

    @ Nick:
    There’s a serious and huge Human Engineering Design Interface issue with the decision for USB3 to be plug-compatible with USB2. Specifically, the loss of the optical connector anywhere in your chain will cause the system to fall back to USB2 speeds.
    For example, a USB3 motherboard and USB3 peripheral, using a USB2 cable …yiedls only USB2 speeds.
    Similarly:
    USB3 PC — USB3 cable — USB2 hub – USB3 cable – USB3 peripheral … yields USB2 speeds.
    This situation (and it has already been committed to) is going to be confusing and highly frustrating to consumers.
    The other thing that I’m concerned with is the ongoing “over-selling” of USB by talking about the 4.8Gbps peak without also mentioning what the protocol’s overhead is, because what’s pragmatically important is the real world throughput (bandwidth). For example, the current situation of USB2 (480) vs FW400, Firewire has ~35% superior sustained throughput, which indicates that USB has at least a 25% overhead. Casual web searches suggest that USB may be as high as 40% overhead. That’s horribly inefficient. I’ve seen some comments on USB3 that the overhead issue is going to be improved, but I’ve not come across any clear, unambiguous benchmarks or even published goals. And it should hardly come as a surprise that some enterprising individual won’t do this and publish it on his own in the first 3 months after both new standards become available. Better start “re-educating” people now, before they get their expectations too far out of whack.
    -hh

  26. Herb says:

    Nick,
    I am trying to do some market research on USB 3.0 and would appreciate any direction you can point me in.
    Do you expect that USB 3.0 will displace USB 2.0 or complement it? It would seem that only a small portion of USB ports would need the higher speeds offered by 3.0. Do you anticipate that in 2012, for example, that PCs would have multiple USB 2.0 ports and only 1 or 2 USB 3.0 ports for high speed transport applications?
    Also, do you expect that USB 3.0 will be positioned as a potential competitor for HDMI?
    Who are the major silicon manufacturers for USB 2.0 and do you think they will be able to transition to 3.0 with its much higher speeds easily?

  27. dpass says:

    get rid of the overhead already!
    i have beter uses for my procss bwidth.
    firewire would have been nice if you guys were not tying to bury it. ieee anyone?