Hands-on Look At USB 3.0, Spec Details Revealed

notdagreatbrain writes "Maximum PC dug up some new information about USB 3.0, got their hands on the new connectors, and even took a look inside the new cables. They learned several new details about the next-gen version of the ubiquitous interface. USB Superspeed will be backward compatible with USB 2.0. The maximum speed of the new spec is 4.8Gbps, which is ten times faster than hi-speed. Five new wires are bundled in the cable, four of them used for data transfer (bi-directional transfer is now supported). More power will also be funneled through the line, so you can charge more devices, faster. The wireless USB is also getting upgraded to version 1.1, and will include ultra-wide band frequency support and Near Field Communication for near-instant swipe-based syncing."

Comments

Full speed, high speed, superspeed

[martin-k]

After USB full speed and USB high speed, we now have USB superspeed. What comes next? Hyperspeed? FTL-speed?

Gotta love the marketing hyperbole...

Re:Full speed, high speed, superspeed

[dotancohen]

After USB full speed and USB high speed, we now have USB superspeed. What comes next? Hyperspeed? FTL-speed?

Ridiculous speed.

Re:Full speed, high speed, superspeed

[HungryHobo]

Unnecessary speed!

Re:Full speed, high speed, superspeed

[Saint Stephen]

ob Ren Stimpy...

"prepare to surge to sub-light speed!.... en.... gage!!!!!!!!!!!"

space............ madnesssssssss.......

Re:Full speed, high speed, superspeed

[moriya]

No no, Ridiculous speed is too slow. We're going... Ludicrous speed!

Re:Full speed, high speed, superspeed

[antifoidulus]

Ludicrous speed is great, save for the gaudy plaid cables that are required.

Re:Full speed, high speed, superspeed

[Mishra100]

What? Nobody has said

WARP speed.

This is slashdot, this speed should be a given.

Re:Full speed, high speed, superspeed

[Ramze]

Ludicrous speed!

Then they go the plaid!

http://www.youtube.com/watch?v=HB7tc9pVvYg

(link to Spaceballs clip on YouTube)

Re:Full speed, high speed, superspeed

[Tweenk]

Lint Speed of course. But we'll need someone extremely scared to go this fast. Maybe each USB 4.0 device will incorporate a clone of Arthur.

Re:Full speed, high speed, superspeed

[Dekker3D]

NEED FOR SPEED!!! mwahahaaaah!
whut? it's a good one.

Re:Full speed, high speed, superspeed

[pmbasehore]

Personally, I'm surprised no one has said

LUDICROUS SPEED!

Like Mishra100 said, this being slashdot, it should have been a given.

I'm just going to wait...

[bucklesl]

...for ludicrous speed.

Re:I'm just going to wait...

[Daimanta]

After that comes "holy shit". I'd love to see that one personally.

Re:I'm just going to wait...

[smoker2]

just follow the pope into the woods ...

Re:I'm just going to wait...

[Tsar]

Some bears are Catholic, but most are Lutheran, owing less to doctrinal agreement than to a species-wide appreciation for large padded pews and good lutefisk.

Armored bears are atheists, of course.

What I want to know

[HungryHobo]

Yes this is great but have any controls along the lines of "trusted computing" been slipped in to these devices. I ask only because it seems to be the fashion now days to try to put as many controls into new technology as possible.

Re:What I want to know

[dotancohen]

Yes this is great but have any controls along the lines of "trusted computing" been slipped in to these devices. I ask only because it seems to be the fashion now days to try to put as many controls into new technology as possible.

Why don't you write to the bodies involved with the development and ask them? If we as consumers don't display our wariness, then why shouldn't the engineers put the "controls" in?

Re:What I want to know

[HungryHobo]

I go the easier route. I just don't buy them.
Plus slashdot has so many engineers reading it that they're more likely to see it than if I email some companies customer care department.

Re:What I want to know

[dotancohen]

I go the easier route. I just don't buy them.

Really? I'd like to see you NOT buy then new 512GB Disk On Key when it comes out in three years because it uses the USB 3 spec, which may or may not contain content controls.

Seriously, stop relying on the engineers to come to you, and start writing to them. The same thing goes for Linux software support: if you want Solidworks to run on Linux, then write to the company and tell them that!

It's going to break.

[dotancohen]

I've seen too many people destroy USB 1 and 2 connectors by repeatedly wiggling the plug out of the sockets to the point where the sockets no longer hold the connector anymore. Now, USB 3 is going to be even deeper, providing even more leverage to ruin the socket with.

Tip: you can repair the USB 1 and 2 socktet by opening the case, placing a thin, flat object on the OUTSIDE on the socket, and giving the object a light tap. Just enough to bend it slightly inward again. Master this skill before USB 3 becomes mainstream.

Re:It's going to break.

[hey!]

This could be fixed if they simply specified the minimum mechanical strength of the sockets and plugs.

A lot of USB cables and devices have connectors you can bend with finger pressure. That's Ok for things like printers that are unplugged once in a blue moon, but it's not good for things like cameras that are frequently connected and disconnected.

Re:It's going to break.

[British]

When I play a game that uses my SOCOM headset, I have to plug it in otherwise ALL audio routes through it. In order to save wear & tear on the USB port on my case, I just use one of those 7-inch USB extension cables. If I wear that out, no big deal.

Just wish Windows would let ME turn off & on the headset in software.

Re:It's going to break.

[NormalVisual]

I've seen too many people destroy USB 1 and 2 connectors by repeatedly wiggling the plug out of the sockets to the point where the sockets no longer hold the connector anymore

IMO it's because the standard specifies a crappy connector with almost no redeeming values mechanically, other than being easy to plug/unplug. They're practically guaranteed to work themselves loose unless the connection is absolutely left alone. There really needs to be some kind of easy, cheap locking mechanism on par with the modular RJ-45 plugs to securely hold the connector in place and prevent the stresses you're talking about. I've had to deal with USB several times in an industrial automation context, and I just hate the connectors. The only decent implementation I've seen is on the cheapest Cognex vision-system cameras, which use a mini-USB connector within a screw-on fixture that locks the connector into place on the camera body and prevents any movement.

Re:It's going to break.

[Reece400]

For my desktop I always make a habit of using usb ports on a PCI card for devices that I plug in & out often so that when the connectors become damages I can cheaply & easily replace the card.

Re:It's going to break.

[Reece400]

Using an external USB hub would also work if the bandwidth bottleneck isn't an issue.

Re:It's going to break.

[dotancohen]

RJ-* is great for "plug it in and leave it alone" situations, which it was designed for. However, it is terrible for connections that must be made and broken on a regular basis, which is what USB was designed for. In particular, the plastic locking mechanism is very fragile and prone to snapping off. I'd say that I've seen that more often than even loose USB sockets.

Re:It's going to break.

[dotancohen]

Any ideas for how they could make the sockets more durable?

Require the socket to be made of a thicker gauge of steel.

USB "Superspeed"

[Apple Acolyte]

Did an Intel marketing manager get the name superspeed from his or her 4 year-old? Couple that with Core i7 and you've got. . . pretty crappy names. I guess Intel's naming schemes have historically stunk (mostly). Here are my suggestions for USB 4, 5 and 6:

USB Superduperspeed
USB Ubersuperduper
USB Ubersuperdupercalifragalisticexpialdocious

Re:USB "Superspeed"

[Andrzej Sawicki]

Did an Intel marketing manager get the name superspeed from his or her 4 year-old?

Well, they are also bringing back the Turbo button, so who knows.

and Yet...

[Lumpy]

It will still be slower for sustained transfers than Firewire 400.

The most important part, did they finally make it non CPU intensive?

I also really want to know what they are targeting with it. as Portable storage has esata which will kick it's butt, and USB2.0 is fine for everything else except video, and we have that standardized on firewire.

Re:and Yet...

[jimicus]

It will still be slower for sustained transfers than Firewire 400.

The most important part, did they finally make it non CPU intensive?

I doubt it. In order to do that, you'd have to move work out of drivers and into silicon, which is quite a bit more expensive.

Re:and Yet...

[sam0737]

Replacing the HDMI, DisplayPort...? Oh nevermind, they didn't enforce encryption on the wire, that's probably not what they are targetting.

Or else Intel would probably get sued by name-not-to-be-mentioned.

Re:and Yet...

[Lumpy]

Yes I have and Firewire and eSATA work fantastically, eSATA kicking everything butt hard in speed.

so again, what are they targeting USB3.0 for? eSata will be the standard by the time they get around to releasing it. I even have a eSata port on my laptop and it's a year old!

Re:and Yet...

[Lumpy]

Actually HDTV camcorders dont have Firewire. HDV camcorders do.

real HDTV camcorders have too much bandwidth to use firewire 400 for transfer and record to a medium that can be read in the computer (Redone uses an array of CF cards, pro cameras use a different system)

The HDV camcorders record low grade Mpeg4 in the same bandwidth that a DV SDTV camcorder uses. It's by loose definition HD by resolution, but the artifacting and quality is so low it's only good for home use.

I use a Canon HD-G1 $5800 "HD" camcorder.. It's not HD by my definition even though it records 1440X1080i. (1/2 HD is what ALL of them record except for the jvc HD7... yes I have one of those as well)

Re:and Yet...

[TheRaven64]

Uncompressed 720p only needs 79MB/s to transfer. FireWire 400 can get close to 50MB/s in real-world use, so it's certainly possible to stream lossless 720p over FireWire 400. FireWire 800 is fast enough for uncompressed video at this resolution.

I've only used SD cameras, but they tend to use a variant of MJPEG, so each frame is losslessly compressed but there is no interframe compression. Using something like MJPEG-2000, you can easily stream 1080p over FireWire 800.

Re:and Yet...

[Hoplite3]

They're targeting everyone with USB 2.0 ports now. People with laptops (which is becoming the largest segment of computer users) have USB ports. If they can buy their next external drive as USB 3.0, they can plug it into the computer they currently have, and when they upgrade to a new laptop with USB 3.0, they'll have an instant speed boost. The power-saving nature of 3.0 will also make it attractive to laptop manufacturers looking to boost battery time. Also, once USB 3.0 controllers end up in the major vendor stack of chips, it'll be hard NOT to get it, just like it's tough to buy a computer without sound (or even superIO).

USB isn't the fastest or least CPU-intensive, but it it by far the most pervasive hook-up on computers. The fact that the same port hooks up everything from humping dog toys to harddrives makes it difficult to knock out of the market.

I think there's probably going to be a bit of a fight between esata and firewire, though. Those seem to be in the same niche -- high-speed data transfer for video et al. Pros and hobbyists will determine the winner there. I still think firewire has more going for it (chaining and a nicer connector), but it's more expensive than esata (or so I heard).

But wireless USB ... yeah. I don't get that at all. USB is all about one connector with backwards compatibility. Take away the connector and ... what's left? I guess if it is cheaper than bluetooth, it might end up in the market.

Re:and Yet...

[A nonymous Coward]

The most important part, did they finally make it non CPU intensive?

Yes. It is interrupt driven rather than polled. Polling was one of the lamest decisions the original USB designers made. For those who don't know the difference, interrupt driven is similar to a phone ringing to get your attention. If it were a polling device, you'd have to pull it out of your pocket every few seconds to see if anyone was calling.

Re:and Yet...

[Jorophose]

it's more expensive than esata (or so I heard

eSATA ports are very cheap to make, especially on external enclosures, which is why they often come with eSATA and USB and no firewire; FireWire would require an actual controller, USB just requires a change of plugs, and eSATA is SATA with a different connector.

Re:and Yet...

[A nonymous Coward]

It IS funny how things are changing so fast. I got out of the Navy to find modular phone jacks had become widespread in just four years. "What are these new fangled things?" Heck, I was used to opening up the "borrowed" ATT phone to silence the mechanical bell so ATT couldn't tell you had an extension in the house by calling and seeing how much power was drawn, and now you could get your own phones?

It is also funny to watch movies around the time cell phones started to become available but were not yet common, so the movie uses one but they have to insert careful dialog to explain it to the audience. "This is my portable phone number." "Is that your cellular phone ringing?"

Then try to explain to some kid why Sergeant Pepper ends the way it does on the flip side, because it made sense on vinyl LPs to take advantage of the final revolution of the spiral joining itself.

Linux and Mac already support USB 3.0?

[dotancohen]

From TFA:
Also, new Mass Storage Device drivers will have to be developed for Windows to take advantage of the spec.

Either Mac, Linux, Solaris, the BSDs and Symbian already support USB 3.0, or somebody at MaximumPC needs to pull their head out from under Ballmer's ballsack.

Wireless USB? Huh?

[Manip]

Could someone please explain the point of Wireless USB to me?

I mean we have WiFi (802.11) for the longer range stuff and Bluetooth for close proximity devices...

What niche does Wireless USB fit in that the existing technology doesn't?

Re:Wireless USB? Huh?

[HungryHobo]

It's for when you want to get something off your pen drive and don't want to bother plugging it in!
Or when someone standing beside you wants to get something off your pen drive without bothering to plug it in!

Re:Wireless USB? Huh?

[sam0737]

WiFi is a general networking thing, and USB remains to be a point to point between host and one devices thing. Use is also device protocol (use of system drivers allow much greater user experience easier), instead of being just a data transmission protocol as in Wifi.
The software (and the user interface design) is much simpler with that...

Like those nasty DHCP, DNS and related failure/exception cases are out of picture. Encryption is also much easier/cleaner to design because the data are never flow between devices.

WUSB is pretty much the same as Bluetooth, just much faster.

while there are not many type of devices that I can think of could make use of 480Mbps without a power adapter, but think of bluetooth which is dead slow, I would love to have my PDA and Camera sync to the desktop over WUSB.

Re:Wireless USB? Huh?

[hey!]

To make money for the vendor who controls the specification and owns key patents?

Seriously, Wireless USB seems to be pretty much a direct competitor to Bluetooth. It is faster than the current generation of Bluetooth but no faster than the next version of Bluetooth as planned. I get the impression it is intended to be simpler than Bluetooth. Bluetooth as service discovery and connection security features that are practical stumbling blocks for average users. Aside from weaknesses in its protocols, the biggest weakness in Bluetooth security that users find it inconvenient. They often leave their devices in insecure configurations and vendors often deliver devices with trivial passkeys like 1234.

WUSB claims to implement security in a simpler way, and intuition tells me that there must be a better way, but still I'll believe it when I see it.

WUSB apparently doesn't have service discovery or security. This clearly makes it more of a bona fide "cable replacement" and certainly simplifies managing the WUSB pairing. This will certainly make connecting devices like cameras simpler; on the other hand it will be up to device driver and operating system developers to figure out how to handle devices that offer an array of services, such as phones. So a lot of application interface standardization goes out the window. That's too bad, although it's questionable whether users currently benefit that much from that.

Although these are significant differences, I'm not sure that they'll be decisive in favor of one technology or the other. I'm betting that everything will depend on how cheap throwing a WUSB interface in a device is compared to throwing a Bluetooth interface in. If one interface costs a nickel and the other costs a dime, I'd bet on the nickel interface.

Full speed

[slashqwerty]

The maximum speed of the new spec is 4.8Gbps, which is ten times faster than hi-speed.

In other news, USB full speed will still be 12Mbps.

Re:It stands for

[gEvil (beta)]

USB stands for Universal Serial Bus. See Wikipedia

Thanks! Do you happen to have an explanation for that IBM thing that's confounded me all these years?

If you haven't got anything good to say about it,

[Joseph_Daniel_Zukige]

iNTEL (wimedia) submarined the xStremeSpectrum/Freescale UWB, which was better tech, just so they could own the patents on all the pipes. That, even though Freescale offered theirs royalty-free.

Now, iNTEL insists on pushing their non-standard UWB into the USB spec.

USB is one of those "We spec our tech conservatively. Our specs are 100% better than you will obtain." technologies. Wireless USB will spill your data into the ether and USB 3, while bursting to n-gigabit, will barely be able to sustain half a gig continuous with only two devices on the line. And multiple bus controllers is an upgrade, still on the drawing board.

Save your money. If serial SCSI is overkill, and your device is not on a LAN, get Firewire. Buy printers with ethernet connectors.

Use USB for keyboards and mice and maybe scanners, like it was intended in the first place.

iNTEL bites.

USB1 and 2 (and now3) = bad connector design

[atari2600]

If I had a penny for each time I inserted an USB cable incorrectly, reversed it and probably managed to insert it correctly, I'd be able to afford one of those high end Mac notebooks.

Looks like the saga will continue with USB3 as the connectors are designed the same. Why can't the connector be designed in such a way that just inserting *would just work* without having to worry about alignment. Too much to ask?

On the flip side, Tannenbaum would be happy: Never underestimate the bandwidth of a station wagon full of 1TB external drives with USB3 ports hurtling down the highway. Or Ritchie - whoever said that.

Re:USB1 and 2 (and now3) = bad connector design

[smoker2]

what you mean like ethernet, or DVI/HDMI, or svideo, or firewire, or serial, or parallel, or IDE, or sata, or even earthed mains ? Christ, even a simple CD needs to be put in the right way up.
Not everything can be practically wired to a "stereo" jack plug, and even if it were practical, how many times will you jam the wrong thing in the wrong socket, sending +5v the wrong way into an expensive bit of kit ?
Even nature uses specific "sockets" to ensure the correct usage. Sure you can stick food up your ass, but you won't be able to chew it, and it'll just drop out anyway. And as for sex, well there is only one socket that really works as intended.
BTW, you never "inserted a USB cable incorrectly" because, surprise surprise, it wouldn't go in, thereby ensuring that you got it right on the next attempt. You may have "tried" to insert it, but you can do that with any socket arrangement.
I'm interested to hear your alternatives for an idiot proof connector layout. One that doesn't involve "keying" the plug to the socket.

Re:USB1 and 2 (and now3) = bad connector design

[Kelbear]

Which side is up on the slot though?

It can be vertical/horizontal and can oftentimes be out of line of sight when reaching behind PCs, or in the dark under a desk.

Re:USB1 and 2 (and now3) = bad connector design

[chaim79]

I think he is refering to 'keying' the plug/socket, all those you reference are keyed plugs (RJ45 (ethernet), DVI, svideo, firewire, etc) are all keyed plugs where the key is visible and easy to distinguish, you don't have to be looking directly at one of those plugs to figure out where the key is, and usually you can see the keying for the socket.

USB decides to be keyed as well, but internally keyed (the key is the shape of the inside of the plug) so you are forced to put the USB logo on one side so people know which side's up. For those who have cables that don't have that label or sockets that are on their side, and you don't know which side is up, this becomes a problem.

I'm seeing people trying to explain that USB is not keyed, but then say it won't go in the wrong way, but that means it's keyed, so you contradict yourself.

Because USB IS KEYED, and it is keyed internally, it's an annoying spec, personally I like firewire better and try and use it more then USB, it's easier for me to see the keying.

Re:USB1 and 2 (and now3) = bad connector design

[TheRaven64]

S-video is a bad design. Serial and parallel presumably refer to D-type connectors (although both come in DIN and Centronics connectors too). FireWire and D-type connectors can have their orientation easily distinguished by glancing at the connector. So can Ethernet. I can't remember what earthed mains looks like in the USA, but I've never seen someone try to plug in a UK mains cable upside down.

With USB, you can only tell which way up it should go by either inspecting both ends closely, or by trying both ways. It also seems very easy to have a USB connector the right way up but not have it slide in easily, not push too hard because you don't want to damage the socket, and just turn it around and try the other way. USB is the only socket I've ever seen where it's common for people to take 3-4 tries to get it right.

Not everything can be practically wired to a "stereo" jack plug

Stereo audio jacks need a left channel, a right channel, and a return. Some use separate returns for each channel to reduce noise. USB connectors have four wires. There is no reason why you couldn't use a 3.5mm four-wire stereo connector for USB, although you'd probably want to use something slightly bigger so you didn't plug it into an audio jack by mistake. There are large numbers of superior connector designs. Raskin devotes an entire chapter to them in The Humane Interface.

Re:It stands for

[gbjbaanb]

lol. but - read the summary: Five new wires are bundled in the cable, four of them used for data transfer (bi-directional transfer is now supported)

So its really a UPB now :)

Re:4.8 Gbps is fast?

[Hal_Porter]

Firewire 800 is how old, and is how fast? About 6.25 Gbps?

If only they named Firewire standards in a way that let users tell how fast it was just from the name.

Yogurt brand cables

[Count_Froggy]

Don't forget the Yogurt brand cables and other devices.

May the Schwartz be with you!

Re:It stands for

[PapaBoojum]

Bidirectional != parallel

Re:Will it have better latency for CNC machines

[Areyoukiddingme]

What are they using instead? We use RS-232 to control a real time external device and under Windows latency is a minimum of 80ms. RS-232 is about as barebones as you can get, so I'm curious as to what else they could be using that's any better?

We need an all new standard

[Skapare]

We need an all new standard. It will probably borrow more from Firewire than USB.

One of the things I would like to see in this new standard is fixed addressing. That is, each port has a number, and the fixed address of the device has that number in an address chain. When you connect devices through a hub, you use a longer address chain. This should be allowed up to at least 8 address units long (as many as 7 hubs to reach a device). Devices will still also have a unique device ID so if a device is moved to a different port it can announce itself there and be found. There is no need to assign addresses to devices at the hardware level. If the OS wants to alias devices, that's fine.

I think having different connector types at each end of the cable is silly. It should be a completely symmetric system where either end can send messages to the other. Then you can connect two computers to each other without needing special hardware in between. Cabling will be simpler (except for tiny devices that need special tiny connectors). An androgynous connector would also be a plus (connect two cables together) and this can be done while keeping bi-directional data paths correctly connected.