USB 3.0 Getting a Speed Boost To 10 Gbps

cylonlover writes "The USB 3.0 Promoter Group has used CES 2013 to announce an enhancement to the USB 3.0 (aka SuperSpeed USB) standard that will see the throughput performance of USB 3.0 double from 5 Gbps to 10 Gbps. The speed boost will come courtesy of enhanced USB connectors and cables that are fully backward compatible with existing USB 3.0 and USB 2.0 devices. The 10 Gbps SuperSpeed USB update (pdf) is up for industry review during the first quarter of 2013, with completion of the standard expected by the middle of the year."

Comments

My Porn

[Anonymous Coward]

I can transfer my porn to thumb drive in only a couple of hours now!!!!

Re:My Porn

[eviljolly]

Don't be ridiculous. You can't fit it all on a thumb drive.

Re:My Porn

[NettiWelho]

You can if its one of these drives.

Re:My Porn

[ArcadeMan]

Correction: you can transfer it all to that magical drive.

Re:My Porn

[GodfatherofSoul]

Hours? Maybe you need a palm drive?

Re:My Porn

[hobarrera]

I'm pretty sure he can.

Standards

[MachineShedFred]

So what's the point of having a version number on your standard, if you don't increment the number when you change the standard?

Customer: "This computer has USB 3, but my 10Gbps device only connects at 5Gbps!"
Support Tech: "Oh, that's because you have USB SuperSpeed 3.0 Revision 1 rather than USB SuperSpeed 3.0 Revision 2."

Maybe call it USB SuperSpeed 3.1?

Re:Standards

[LWATCDR]

So USB could replace SATA. Well if they get the overhead down.
I would really like to see a SATA IV spec that is a little faster but includes power on the connector. It makes little sense to me to have separate connectors for power and data on SATA since you can not have an unpowered SATA device.

Re:Standards

[Scutter]

It doesn't matter. Most of the time, the hardware manufacturers don't print the rated speeds on the cables (and usually don't even print the standard - USB 2.0, USB 3.0, etc.). They just use their own marketing lingo that generally has no bearing on the standard whatsoever. It's difficult (at best) to compare apples-to-apples when buying a simple USB cable.

Re:Standards

[drinkypoo]

On one hand, USB already confused consumers by supporting a plethora of speeds. On the other hand, today SATA is sold by speed and that doesn't seem too confusing for people.

Re:Standards

[Pieroxy]

It is a trademark of USB. Not one USB standard has had a single speed nor has it had its speed easily recognizable from the marketing garbage spilled by the consortium. And I'm not even talking about the mess of mixing USB1 & 2 devices and hubs. USB Full Speed, Hi Speed, Low Speed... and now SuperSpeed.

To illustrate, here is an excerpt of the Wikipedia page:

High-speed USB 2.0 hubs contain devices called transaction translators that convert between high-speed USB 2.0 buses and full and low speed buses. When a high-speed USB 2.0 hub is plugged into a high-speed USB host or hub, it will operate in high-speed mode. The USB hub will then either use one transaction translator per hub to create a full/low-speed bus that is routed to all full and low speed devices on the hub, or will use one transaction translator per port to create an isolated full/low-speed bus per port on the hub.

Garbage.

They obviously HAD to do the same for USB3, for old times' sake. We will laugh about it to our grandchildren next to the fireplace. But that'll be later.

Re:Standards

[LordLimecat]

USB will never replace SATA:
  * It hits the CPU for each transfer
  * the overhead is higher
  * The latency is way higher, as it needs to set up and tear down connections for each transfer
  * It doesnt support ATA commands (TRIM, for one)

Re:Standards

[Anonymous Coward]

quibble: the usb mass storage spec does allow for ata,
even if it is very seldom used.

Re:Standards

[craigminah]

No kidding. I bought a college textbook, it was "5th edition" and just released prior to my semester starting. After the semester I tried to sell it only to be told by the college bookstore that it had been replaced. I looked and the book that replaced it was "5th edition" then the bookstore clerk said it was "5th edition, 2nd revision". My response probably made me sound like a cross between Joe Pesci and Yosemite Sam. Stupid book publishers...

They should at least change it to USB 3.1 or something, like what they did with HDMI versioning.

Re:Standards

[wonkey_monkey]

I would really like to see a SATA IV spec that is a little faster but includes power on the connector.

Isn't that what eSATAp is for?

Re:Standards

[Anonymous Coward]

Maybe call it USB SuperSpeed 3.1?

It will never take off then. Everyone will wait for USB 3.11 For Workgroups.
 

Re:Standards

[LWATCDR]

I guess you missed the comment about getting the overhead down.

Re:Standards

[Anonymous Coward]

Just stay away from USB LudicrousSpeed. I tried it and my wallpaper was changed to this

Re:Standards

[operagost]

Half.com is your friend.

Re:Standards

[LWATCDR]

That seems to me what SATA should have been.

Re:Standards

[Anonymous Coward]

Ok so yeah maybe if you took USB and removed everything that makes it USB then USB could replace SATA.

Idiot.

Re:Standards

[Svenne]

They won't do that because then consumers will know the difference, and as it turns out, that's bad for business.
 
Remember http://hardware.slashdot.org/story/03/06/18/2025210/usb-11-renumbered-to-usb-2?

Re:Standards

[Immerman]

Simple explanation: SATA is primarily an internal-facing interface. "Consumers" almost never see it, only tech-savy individuals. And even the slowest SATA standard is still drastically faster than all but the most cutting-edge drives can use so in most cases the version doesn't actually matter much anyway.

USB on the other hand was specifically designed as a user-facing interface to make life easier for people who had trouble getting all the different "can only fit in one socket" cables plugged into their computers*. Well, in addition to solving all the limited IRQ headaches associated with using multiple serial and parallel ports. And allowing easy hub-based fan-out.

*never quite understood that - it's not *that* much harder than the blocks-and-holes puzzles they solved as a toddler. Well, except the identical but non-interchangable PS-2 mouse and keyboard plugs - those were always a headache if you couldn't see the labels.

Re:Standards

[Anonymous Coward]

I would like to buy a vowel please.

Re:Standards

[drinkypoo]

*never quite understood that - it's not *that* much harder than the blocks-and-holes puzzles they solved as a toddler. Well, except the identical but non-interchangable PS-2 mouse and keyboard plugs - those were always a headache if you couldn't see the labels.

"Back in the day" it was not unusual to look at the back of a PC and see two DIN plugs and three or even more unlabeled DB-9 plugs, each of which had different signalling requirements and might let out smoke if you get them wrong. And in fact, it was not unusual to look at the back of a Mac and see two DB-15s and two DIN-4s, again with completely different signalling and power. So it's only today that I find it particularly inexplicable when people can't set up a computer when things literally work anywhere you can plug them in. My motherboard doesn't even care where I put my RAM.

Re:Standards

[ACluk90]

This seems so wrong to me...
  * current controllers hit the CPU for each transfer
        -> there will be hardware acceleration if this is wanted
  * the overhead is higher
        -> what overhead? data overhead which is so much trouble if we have such an excessive amount of bandwidth?
  * The latency is way higher, as it needs to set up and tear down connections for each transfer
        -> the connection does not need to be opened and closed for each transfer - why not keep it open?
  * It doesnt support ATA commands (TRIM, for one)
        -> any kind of data can be transmitted over USB, also a TRIM command; but I have to admit that there might be need for a new 'mass storage class' that is targetted at SSDs.

Re:Standards

[Anonymous Coward]

if it worked like that, it would no longer be usb.

However.. neither does usb 3.0!

Re:Standards

[Culture20]

I would really like to see a SATA IV spec that is a little faster but includes power on the connector. It makes little sense to me to have separate connectors for power and data on SATA since you can not have an unpowered SATA device.

I would rather not have power for 5 or more SATA drives running through my motherboard just to make a unified connector the standard. Same goes for RAID cards (especially the 8+ port ones).

Re:Standards

[LordLimecat]

You cannot get the overhead down without making USB into a storage-specific protocol...
at which point you've just re-made eSATA.

Why not just use the storage specific protocol we already have?

Re:Standards

[LordLimecat]

-> what overhead? data overhead which is so much trouble if we have such an excessive amount of bandwidth?

The overhead that comes from using a non-point-to-point, general purpose protocol.

-> the connection does not need to be opened and closed for each transfer - why not keep it open?

Not being a USB engineer, I cannot answer that, but I have a feeling it has to do with the fact that you could in theory have a hub connected and a printer might want use of the USB line for a bit, and an "always open" connection would block that. Its just a guess.

But regardless, throw an SSD on USB 3 and run IOmeter, then compare with an SSD on SATA. The USB one will have latency ~10x higher or worse, potentially much worse. response times for SATA will be under 1ms (closer to 0.5ms), response times on USB can be 10+ ms. I believe I saw as high as 40ms in my testing. Its been about 15 years since USB came out, if what you were suggesting were a no-brainer I think they would have done it already.

-> any kind of data can be transmitted over USB, also a TRIM command; but I have to admit that there might be need for a new 'mass storage class' that is targetted at SSDs.

They already have something like that, its called eSATA. USB isnt a storage protocol, and trying to make it one will just make it a poor copy of SATA.

Re:Standards

[Anonymous Coward]

USB solves a very broad domain of problems, any optimizations that make it remotely competitive with other protocols like SATA would make it no longer good for its original problem set.

USB is like RESTFUL HTTP. Decent performance, easy to use, but if you need more performance, you need another protocol. It's a "jack of all trades, master of none" protocol.

Re:Standards

[skids]

Maybe call it USB SuperSpeed 3.1?

I vote for "USB_maybe_we_should_have_just_used_UTP_RJ45s_in_the_first_place."

Re:Standards

[Espectr0]

Maybe call it USB SuperSpeed 3.1?

I am waiting for USB SuperSpeed 3.11 for Workgroups.

Re:Standards

[sjames]

You're close. It's not a 'connection' so much as it is a token allowing the target device to talk. It can't just be left with one target in case another target might need service.

So being non p-t-p is highly relevant as you guessed.

Re:Standards

[sjames]

That must have been devastating for your tennis game. Mixing metaphors is fun!

Re:Standards

[Immerman]

Really? Maybe I'm blocking something out, but most every personal computer (servers, etc were admittedly a whole different ballgame) I worked on had very limited connectivity options - PCs had their keyboard DIN (or later the ps-2 mini-DIN pair) rarely more than two DB-25 or -9 serial ports, and a DB-25 parallel port with an opposite gender, and the VGA port with an extra row of pins. Oh, and often a game port, can't forget that :-). Nothing that could be plugged in wrong. Seems like one of the early SCSIs used a DB-25 as well, but that wasn't seen much outside the high-end systems and Macs, and Macs, well okay, Macs were a little over-obsessed with their pretty matching mini-DINS and such. Seems like most of them still had different pin layouts so they couldn't be plugged in wrong without forcing though, it just made it pretty much impossible to find the right port by feel.

Yeah, it's certainly even more inexplicable today though. It's like some people just see a bunch of wires and their brains just say "uh-oh, complicated" and shut off completely. Then again maybe it's the very universality that throws them for a loop - "All the plugs are the same! How do I tell where they're supposed to go?!?" Certainly an understandable response given the fact that the rest of the world is still liberally laced with mechanically compatible plugs that will do bad things if connected.

Re:Standards

[davester666]

USB SuperDuperSpeed 3.1!

Re:Standards

[Cytotoxic]

I remember, DrinkyPoo.... Oh boy, do I remember. When I was a kid doing a lab rotation I made the mistake of plugging what I thought was a serial line printer that we had laying around in a store room into a PC serial port to replace the broken line printer for the gamma counter. Ooops. Everything looked fine for a few seconds, then the printer started spewing paper and printing random gibberish. Then it let the blue smoke out. Fried the printer's board. I think the printer might have gone with an old VAX that had been replaced. I dunno. It was a dumb enough mistake that I took my lumps without asking any further questions. The lab manager, techs and grad students who helped with the fiasco were grateful enough that I took the sole responsibility that we never spoke of it again. But I bet none of us ever made that "the connector looks the same so it must work" mistake again.

Re:Standards

[nabsltd]

I would rather not have power for 5 or more SATA drives running through my motherboard just to make a unified connector the standard. Same goes for RAID cards (especially the 8+ port ones).

Or SAS/SATA port expanders, which currently only need a very small amount of power (as they are just re-routing the low-power data signals), but with built-in power, you could even go beyond the 75W limit for graphic card draw from the slot, never mind the 35W "other" card limit.

Re:Standards

[Culture20]

It's difficult (at best) to compare apples-to-apples

Sometimes it's difficult to compare Apples to Apples.
"I'd like to Max out my RAM. I have an iMac"
"Sure, but what version? That makes all the difference for what RAM it can use"
"Um, it's got a black back and I bought it in 2011 from a friend who upgraded to a newer Mac. I don't see a version number."
"Call your friend and find out when it was purchased from Apple"

Re:Standards

[LordLimecat]

From what I've read, the connection set-up (handshakes, acks, etc) is also a large cause of the latency. A device cant just start throwing data down the wire, it needs to negotiate, make sure it was heard, and generally assume that the connection is prone to packet loss; and all of that generally slows the connection down.

Re:Standards

[Cid Highwind]

*never quite understood that - it's not *that* much harder than the blocks-and-holes puzzles they solved as a toddler.

Users have experienced the USB disorientability principle (try to plug in USB cable, doesn't fit, rotate connector 180 degrees, still doesn't fit, rotate connector again, plugs right in) enough times that they're concluded computer cables exist in some inscrutable 5-dimensional space and given up trying to understand what plugs in where and why.

Re:Standards

[Culture20]

I'm guessing you've never seen people put USB into RJ-45 ports (I've even done it on a computer where I couldn't see the back and the NIC port was right next to the USB). It fits perfectly on a lot of systems.

Re:Standards

[Anonymous Coward]

Because nobody wants another external port. Does it really have to be spelled out for you? It's not like we have a huge supply of eSATA products or laptops with such a stupid port.

Many external drives don't even use a SATA bridge chip. They only understand USB.

Oh, and it does support TRIM through the new USB SCSI protocol.

Re: Standards

[Xeranar]

I thought I was the only one! Same situation, couldn't see the back and instead of using my finger I used the plug and inserted. About 15 minutes of me thinking my mouse was dead was rectified when I pulled it from its wedge and found it in the extra ethernet port. >.

Re:Standards

[Immerman]

Hoo yeah - once you get into the world of non-consumer hardware there was (and still is) a LOT of potential for magic smoke release - after all the D-sub plugs are industry standard mechanical descriptions for generic electronic interconnects to be used as-needed. I remember the physics labs being *full* of devices that you absolutely, positively did not want to try plugging into a computer, no matter how temptingly compatible the plug looked. That D-subs were used for PCs at all is likely a cost-cutting decision made in the early days. Well that and the fact that idiot-proofing probably wasn't taken too seriously with what were initially extremely expensive decidedly non-consumer devices, well beyond gender-swapping the serial and parallel ports so you couldn't accidentally plug into the wrong one.

Re:Standards

[Immerman]

Fair enough, I've done it many times myself when plugging things in by touch, especially when they're immediately adjacent to a USB port. I'd argue on the perfect fit though - the width is right, but it's a pretty sloppy fit vertically, and if you're paying attention to what you're doing it only takes a couple seconds thought to realize that the purpose of a plug is to make electrical connections, so the combination can't possibly be right.

Re:Standards

[sjames]

Transactions do add a fair bit of latency as well.It's part of the cost of needing to handle a device being suddenly disconnected while not disrupting other devices in the same tree. Your keyboard needs to keep working even if you unplug a USB stick in mid-write.

Re:Standards

[Culture20]

Yep, that's why I do a four-way wiggle whenever I blind-insert these days.
I can't wait for someone to take this post out of context.

Re:Standards

[teg]

It's difficult (at best) to compare apples-to-apples

Sometimes it's difficult to compare Apples to Apples. "I'd like to Max out my RAM. I have an iMac" "Sure, but what version? That makes all the difference for what RAM it can use" "Um, it's got a black back and I bought it in 2011 from a friend who upgraded to a newer Mac. I don't see a version number." "Call your friend and find out when it was purchased from Apple"

Just select "About this Mac" and then "More info" and you get the information you need - "Macbook Air, 13 inch, Mid 2011" in my case. Plus some extra information about processor, memory (installed size, type and speed) etc.

Re:Standards

[LWATCDR]

Good point but then CPU power is now getting silly cheap. I could see low end systems just using USB for everything including internal mass storage just to save money.

Re:Standards

[Carnildo]

*never quite understood that - it's not *that* much harder than the blocks-and-holes puzzles they solved as a toddler. Well, except the identical but non-interchangable PS-2 mouse and keyboard plugs - those were always a headache if you couldn't see the labels.

The PS/2 mouse, the PS/2 keyboard, and the Zip drive power connector all use a 6-pin mini-DIN plug, and are completely non-interchangeable (plug a mouse into your Zip drive, it doesn't work. Plug a Zip drive power brick into your computer, and watch the sparks fly).

The PC parallel port and the Mac external SCSI port both use female DB-25 connectors; confusingly, parallel-port printers have male Centronics 36-pin ports, so you need a mismatched cable to plug your printer in. Further confusing the issue, Zip drives, scanners, webcams, and just about any other peripheral you can imagine plugs into the parallel port, usually with daisy-chaining that never quite worked properly.

The modem jack and phone passthrough jack are identical, but few modems will work properly if attached backwards. Further, you can plug a phone cord into an RJ-45 Ethernet port, but it won't work.

Mice come in both DB-9 serial and PS/2 mini-DIN format, and speak different protocols, so if you've got an adapter, you also need to make sure the protocol selector switch on the mouse is set correctly. Similarly, keyboards come in both PS/2 mini-DIN and 5-pin DIN format, and speak both the AT and XT protocols.

Serial ports come in four flavors: DB-25 male RS-232, DB-9 male RS-232, DB-9 male RS-422, and 8-pin mini-DIN RS-422 (the latter two seen mainly on Macintosh computers). You can sometimes use passive adapters to switch between them, but there's always the possibility of having an RS-422 device that's picky about the signaling details, or an RS-232 device that needs the extra control lines of the 25-pin connector.

Analog video is usually either female DE-15 or DE-9, but computers made by Apple use female DB-15 -- and the Apple and Macintosh lines use different signaling protocols over the same DB-15 connector, neither of which is compatible with the PC's DE-15 standard.

Older joysticks usually have unshielded female DE-9 plugs (and so can attach to DB-9 serial ports, but tend to let the magic smoke out if so used). Newer joysticks use shielded DA-15 connectors, and if the joystick port is on a sound card, it doubles as a MIDI port. You can attach a Y-cable to your joystick port to plug in two joysticks (assuming both are basic two-axis two-button models), but you can't daisy-chain Y-cables to plug in more, and you can't use it as a splitter to plug both a MIDI device and a joystick in.

Also on the subject of sound cards, your computer probably has three 3.5mm three-connector jacks: two on the back, and one on the CD-ROM drive. One of the jacks on the back is for speakers while the other is for a microphone; the one on the drive lets you use the drive as a CD player by plugging in headphones, and won't play sounds produced by the sound card.

The shielded coaxial connector on the back of your computer? It might be Ethernet, it might be Token Bus, or it might be composite video. Have fun!

Re:Standards

[drinkypoo]

PCs had their keyboard DIN (or later the ps-2 mini-DIN pair) rarely more than two DB-25 or -9 serial ports, and a DB-25 parallel port with an opposite gender, and the VGA port with an extra row of pins

Before VGA, there were MDA, CGA, and EGA, each on a DB9. And many early busmice also used a DB9... but they also used DIN ports. And it was not unusual to get a busmouse and an ISA card to match cheaper than a PS/2 mouse at the time, either. If you had a light pen it was probably proprietary and often on a DB9. And these are just the examples I can come up with off the top of my head.

Amusingly, on Intel motherboards you can or could commonly swap KB and Mouse and they would work fine, but I don't recall any IBM hardware where this would work.

Seems like one of the early SCSIs used a DB-25 as well, but that wasn't seen much outside the high-end systems and Macs, and Macs, well okay, Macs were a little over-obsessed with their pretty matching mini-DINS and such.

Cheap PC SCSI cards commonly used the DB-25 that was popularized by Apple at the time, because it's a cheaper connector than the centronics and the HD50 was only just being invented and amphenol wanted very much money for them. Such garbage cards were commonly bundled with SCSI scanners when sold to the PC market. The DB25 to Centronics cables were the same price as or cheaper than Centronics to Centronics. I have lost track of how many times I had to wiggle a centronics connector through the back of a PC to install an early Adaptec SCSI card.

Macs were a little over-obsessed with their pretty matching mini-DINS and such. Seems like most of them still had different pin layouts so they couldn't be plugged in wrong without forcing though, it just made it pretty much impossible to find the right port by feel.

The problem then came when you installed an Add-in card, because ADB uses the same connector as S-Video. I used to have a TV card with S-Video installed into a Mac IIci... And as stated previously, apple monitors used the same-gender DB15 as an AUI card, which is why Apple invented their lame little AAUI connector. But you could buy NuBus ethernet cards (e.g. from 3Com) with an AUI connector, I have one right here...

Re:Standards

[hobarrera]

Power comes from the PSU, while data goes to the motherboard.
If a single connector carried everything, then power would have to flow through the motherboard. That's a lot of power if you've 8 disks connected to it. Power that would heat up lots of circuits, etc.

Re:Standards

[jnork]

I think what it comes down to is that USB is a general-purpose protocol designed to be a convenient way to connect a large variety (and number) of devices to your computer quickly and easily. And cheaply. While high performance is desireable for mass storage and streaming, USB wasn't designed to be a performance bus, it was designed to be flexible and convenient. Whereas SATA was designed for mass storage. Primary considerations are performance and reliability.

I don't see USB replacing SATA any time soon. Increasing the raw bitrate isn't enough.

Re:Standards

[jnork]

"All spelling and grammar errors are intentional. Grammar Nazis' need entertainment."

You're killing me. You know that, right? :)

Re:Standards

[Anonymous Coward]

Nah, you want USB 3.51.

Re:Standards

[Anonymous Coward]

USB 3.0 doesn't deliver 5 gbps yet... I have many USB 3.0 devices and a brand new MOBO and the most I have gotten out of it is 300 mbps to HDD, and almost 400 mbps to SSD, so far my eSATA port gives me more throughput than my USB 3.0 does. So if it cant deliver 5 gbps yet and doesn't even out-perform eSATA, maybe they should work that out before upping the standard to 10 gbps.

Will this speed ever be used?

[Anonymous Coward]

While I understand that it's a great thing to have such high transfer speeds, I'm beginning to wonder what speed we'll get in USB 4.0 when it eventually appears. And whether the huge investment into the standard will be worth it, given the fact that nothing will ever need to transfer that much data per second.

Re:Will this speed ever be used?

[MiG82au]

LOL, will you ever learn? I remember thinking I'd be happy if my internet were as fast as a floppy disk...

Re:Will this speed ever be used?

[vlm]

I think its a matter of scale. So raw uncompressed video signal on an original displayport connector is 4.32 Gbit/s... now instead of a "USB" port being able to carry one uncompressed video signal it could carry two. Of course displayport 1.2 does 17.28 Gbit/s, so we've gone from only 1/4 to only 1/2 of a video signal. On the other hand, displayport has about twice the BW available as HDMI, so you could just about replace HDMI with USB now on a raw available BW basis. Although it would be a heck of a lot more intelligent to shove compressed video down the cable rather than raw uncompressed.

Using USB for a uncompressed video connection is not a valid or useful data point, anyway. But it does make the point that this is competitive with the fastest port anyone is likely to ever have access to. Nothing in your average dude's computer will ever be able to saturate a USB thats faster or about as fast as the video cable. A more normal use case is I'm sure my typing speed was not limited by 5.0 Gigs USB so 10.0 Gigs USB is not going to help.

In the very long run we will not have USB / Firewire / SATA / PATA / Displayport / HDMI we'll have just one connector and protocol to run them all. Plug your keyboard, mouse, LAN adapter and monitor into your hub connected to your phone and be done with it. The only question is which standard will win. Probably USB.

Re:Will this speed ever be used?

[SJHillman]

Nothing? Have you paid attention to computing in the last 10 years? Everything is interconnected... if monitor resolutions go up, you damned well better believe the average picture size will go up just so it can be viewed fullscreen without getting blurry. Movies will follow a similar trend (think the SD to HD switch). How about games? I remember how amazed I was that Diablo 2 took so many CDs... and now there's games that take multiple DVDs. What about Windows... and installing from USB thumb drive is becoming icnreasingly popular for software... including operating systems. Not to mention actually running software from USB devices. People will not be happy until even the largest install/transfers happen instantly.

And if you look at corporate arenas, USB storage is very popular for backups. The USB 2.0 interface is currently the biggest bottleneck where I work. USB 3.0 would work for now, but our backups are growing by about 2GB per week (we back up nearly 100GB more now than we did a year ago for the same set of jobs) and trend isn't likely to change in the near future.

And let's not forget niche applications, especially military and science, where you measure data in petabytes and exabytes. All that data needs to be moved sooner or later.

One of these days, USB may even rival the bandwidth of a station wagon full of tapes hurtling down the highway.

Re:Will this speed ever be used?

[samkass]

In the very long run we will not have USB / Firewire / SATA / PATA / Displayport / HDMI we'll have just one connector and protocol to run them all. Plug your keyboard, mouse, LAN adapter and monitor into your hub connected to your phone and be done with it. The only question is which standard will win. Probably USB.

That was the hope with Thunderbolt/LightPeak, which is on all Macs these days and works well. One cable carries two full-duplex 10Gb channels (10Gb each way simultaneously per channel). But "docks" have been slow in coming and expensive. And because the USB group refused to integrate the standard or allow them to use the connector, they switched to the DisplayPort interface which is nice and compact. Now we have a slower standard coming much later for which existing cables may or may not work but look the same as the current ones... fun.

Re:Will this speed ever be used?

[Quila]

For lower speed we have Lightning. You get 8 pins plus ground. Over that, you can run pretty much anything that the pins will support, with the embedded chips in the ends doing any pin reassignment. USB3 is 8+ground+shield, so it could probably be run over Lightning. We should be able to run analog audio and video, as well as HDMI and other things too. It is the universal low-speed point-to-point connector.

Thunderbolt is meant for higher purpose than USB3, since you can theoretically run multiple monitors, RAID arrays and even an external video card over one cable at the same time. Basically, you get that old docking station via one small plug.

Of course there will be some meeting in the middle, as USB3, HDMI, Gigabit Ethernet and others can be run on both.

Re:Will this speed ever be used?

[davydagger]

"That was the hope with Thunderbolt/LightPeak, which is on all Macs these days and works well. One cable carries two full-duplex 10Gb channels (10Gb each way simultaneously per channel)"
terribly complicated $50 cables with microcontroller sin the wire to handle connections.

They are basicly PCI-E lanes built into an external jack. Intel wanted to do fibre optics with the protocol, but it never happened.

"Now we have a slower standard coming much later for which existing cables may or may not work but look the same as the current ones... fun."

backwards compatible with devices. thats what is going to sell.

Re:Will this speed ever be used?

[Anonymous Coward]

Intel planning on having the fiber optic version of LP ready around 2014/2015, the fact that a copper version even appeared is strange. Intel expects the fiber version to scale up into the 100Gb range and replace internal PCIe slots.

Re:Will this speed ever be used?

[Rockoon]

The entire problem with Thunderbolt/LightPeak is cost.

The deal is that if you want a generic interface for lots of purposes, its got to be cheap, or many of those purposes simply don't make any sense when a much more economical solution exists that doesnt require any of those "advanced" features that drive up the cost.,

On the other hand if you want a specific interface with a specific purpose, its also doesnt make sense to increase costs with generic requirements.

The upshot is that by the time the market really looks at Thunderbolt/LightPeak as a viable solution outside of economics-ignoring niches (where the device is expected to be expensive anyways), a cheaper solution will be waiting. Here we are with USB getting a bump that reduces the size of the Thunderbolt niche, and then SATA/eSATA will also get a bump, reducing that niche further...

I agree with the many posters that suggest that SATA3 was too small an increment, and I was even saying that before it was finalized because it was quite apparent that SSD's were already interface limited on SATA2. It didnt' take but a few months for interface-limited SATA3 SSD's to hit the market. The next SATA (Express) will be up to 16gbps.

Re: Will this speed ever be used?

[Xeranar]

I've heard this argument before but I fail to see where the benefit is for the average consumer. Most things I would plug into a PCI-E 1/4/8/16 I wouldn't want lying on my desk. There are roughly three common items people would use, Video cards, Sound cards, and wireless adapters. I fail to see the advantage. If we're discussing removable storage eSATA and USB are fine. Thunderbolt is great for thr future but sans video card housing for laptops there is little use for it as a PCI-E port for the forseeable future.

Re: Will this speed ever be used?

[Xeranar]

Time is the constraint on the human condition. We damn well should want mundane tasks to be instanteous.

Re:Will this speed ever be used?

[MiG82au]

How big are 4k high frame rate 3D movie files are going to be? And why would you want to wait longer than you have to? I already hit sequential speeds of 800 MB/s (~8 Gb/s) two years ago with a RAID 0 array of cheap 60 GB SSDs (for fun, not actual usage). Speeds of multiple GB/s are going to be a reality and the average user will find a way to use it. How about off site backups of TB magnitude data? Sure, if you're serious you will use a proper drive interface rather than the hypothetical USB 4, but at some point it will be something that the common person will do, and I can see there being a USB 4 or equivalent to make it easy.

There may be other reasons that USB 3 will be the last USB standard (like your one protocol idea), but the original poster's opinion that "nothing will ever need to transfer that much data per second" is a load of bullshit.

USB 3.1?

[Anonymous Coward]

Can we call it USB 3.1 or something so it's less confusing?

Could we at least make them a different color?

[Anonymous Coward]

Or mark them in *some* way so I can tell all the basically identical USB ports and cables apart?

Re:Could we at least make them a different color?

[SuricouRaven]

USB3 cables have a differently shaped connector from USB1-2.

Re:Could we at least make them a different color?

[Anonymous Coward]

Not really, there are a couple new "B" connectors but the "A" connectors are exactly the same.

Re:Could we at least make them a different color?

[SJHillman]

All of my USB devices have either a description or symbol right on the male plug. The female ends are usually labeled too. It's not as quick and dirty as color coding, but it does work better when you consider the reality of a world where not everyone follows color coding standards.

Re:Could we at least make them a different color?

[SJHillman]

Wikipedia has pictures, but basically says A is the same but B is larger and won't fit in older B slots.

From Wikipedia:

USB 3.0 connectors
- Type A plugs and receptacles from both USB 3.0 and USB 2.0 are designed to interoperate.
- Type B receptacles in USB 3.0 are somewhat larger than would be required for a Type B plug in USB 2.0 and earlier. The larger dimension for a USB 3.0 Type B receptacle is intended to allow connecting of either the larger USB 3.0 Type B plug or the smaller USB 2.0 or earlier Type B plug into a newer USB 3.0 Type B receptacle. Accordingly, a USB 3.0 Type B receptacle on a peripheral device can be connected using the corresponding plug end of a USB 2.0 Type B cable.
- Type B plugs in USB 3.0 are somewhat larger; therefore, a USB 3.0 Type B plug cannot enter a USB 2.0 or earlier Type B receptacle. Accordingly, normal USB 3.0 Type B plugs cannot be inserted into normal USB 2.0 Type B receptacles found on peripheral devices (and connect them to a computer).
- A receptacle for eSATAp (eSATA/USB Combo) is designed to accept USB Type A plugs from USB 2.0 and USB 3.0.

Re:Could we at least make them a different color?

[jessehager]

The new USB 3.0 cables are supposed to have a different logo embossed on them containing an "SS" connected to the old USB "tree" logo. - See section 5.5.6 of the USB 3.0 spec.

USB 3.0 ports and the plastic bit inside the cable connectors are supposed to be color coded blue (Pantone 300C is recommended) - See section 5.3.1.3 of the USB 3.0 spec.

Re:Could we at least make them a different color?

[LeadSongDog]

It'll be easy,
the two connectors will be 50mm apart to minimize transmission losses.

Re:Could we at least make them a different color?

[ericloewe]

The extra contacts are also a dead giveaway

Re:Could we at least make them a different color?

[nabsltd]

Wikipedia has pictures, but basically says A is the same but B is larger and won't fit in older B slots.

This is one of those cases where Wikipedia hasn't kept up with the real world, because I haven't found any USB 3.0 device (which is where the B-connector resides) that uses anything but a micro-B connector, which is completely different from the old B connector.

On the other hand, the A connector/plug backward compatibility works perfectly, as I can plug a USB 2.0 cable into a 3.0 connector on my PC, or a 3.0 cable into a 2.0 connector.

Re:Could we at least make them a different color?

[Anonymous Coward]

I have a USB 3 HD enclosure with the big B-connector. It's made by Vantec.

I can't find cables with the new micro-B connector on monoprice.

Re:Could we at least make them a different color?

[nabsltd]

I can't find cables with the new micro-B connector on monoprice.

Since the device should come with a cable with the micro-B, you can just use an extension with the original cable.

You can also get decent prices at Newegg for full replacement cables. Remember to change the sort to "Lowest Price".

Re:Could we at least make them a different color?

[Anonymous Coward]

The A connectors are NOT "exactly the same". There are extra pins buried deep in the connector that make the extra connections needed for USB 3.

Kinda like EISA slots could support ISA cards, but EISA cards made extra connections deeper in the socket.

The B connectors, on the other hand, put the extra pins in a second section beside the first. They made the extra section narrower so you could only plug a USB 2 micro B plug into one place (the correct place).

As for different colour, the convention is that USB 3 A-type ports have a blue tab, or are labeled with the USB symbol + SS (for SuperSpeed), or both.

can we call it

[alta]

SuperDuperSpeed USB?

Re:can we call it

[bennies]

"Ludicrous speed USB". Like in : http://www.youtube.com/watch?v=ygE01sOhzz0

Re:can we call it

[satch89450]

They've gone into plaid. http://www.youtube.com/watch?v=mk7VWcuVOf0

Re:can we call it

[Anonymous Coward]

Nope, next generation will be called Ultra USB Turbo+

Re:can we call it

[vlm]

Ultra is more 60s, and Turbo is more 80s.
I'm guessing something inspirational/aspirational like "USB Vista" or a made up word like "USB Zune" or add an R at the end with no vowel, so "USBr"

There has been some roman numeral stuff in recent phones, along with astronomical themes, so my best guess at this time for the next USB standard marketing name is "Stellar USBr IV+"

Re:can we call it

[SJHillman]

Other buzzwords we can use:

Mega USB
Giga USB
Jigga USB
Wowza USB
I-Can't-Believe-It's-Not-Instant USB
MS-USB
USB Reloaded
USB The Wrath of Khan
This USB Standard Was Sponsored By McDonalds
Democracy USB
War On Slow Speeds USB
iUSB
USB 3.14159265359

Re:can we call it

[Anonymous Coward]

Not confusing enough, I suggest "Thunderbolt USB". That way, the PC guys get to screw Apple fans over while simultaneously screwing themselves. I think it's called circle jerk or something.

Re:can we call it

[Anonymous Coward]

"USB 3.14159265359" ... Mmmmm ... pie ...

Re:can we call it

[deroby]

You forgot Freedom USB =P

Re:can we call it

[Nivag064]

WOSS-USB : War On Slow Speeds USB

Note 'WOSS-USB' is pronounced 'was usb'.

Also, note that any product starting with an 'i', such as an iPad, indicates that it relies on the imagination of the purchaser to see more value in the product than the price warrants. Remember 'i' stands for the square root of minus one - an imaginary number!

CPU utilization, pc to pc, and Power not speed

[silas_moeckel]

USB3 is fairly fast as it is. It uses far to much CPU time right now as in pegging a cpu while writing 150MBs while the internal sata's on the same machine writing to the same model drive is 20%. The enhanced power is not part of the base standard so there is a chicken and the egg issue with anything using it it needs to be baked in. The USB3 spec allows for pc to pc connects but again it's not a requirement to support it so no OS supports it.

Re:CPU utilization, pc to pc, and Power not speed

[Anonymous Coward]

Worse for me. USB3 pegs my CPU while SMB file transfers at 114MB/s only uses 0.5% cpu and SSD internal transfers of 500MB/s is more along 5%-8%. USB is horrible for higher performance, but it is awesome for what it is meant to do.

Re:CPU utilization, pc to pc, and Power not speed

[Anonymous Coward]

The USB3 spec allows for pc to pc connects but again it's not a requirement to support it so no OS supports it.

You mean just like USB 2? The feature that Linux has supported for ages?

Re:CPU utilization, pc to pc, and Power not speed

[Anonymous Coward]

Hasn't the USB group thought of putting a processor on a USB board, separate from the motherboard, specifically for high efficiency uses?

I mean, yeah, processors are getting better every day and game devs mostly refuse to take advantage of multicore directly, but still, older machines are more the target for such a card anyway.

Or does such a thing already exist?
Seems they exist, but as to whether they have processors specifically to take a lot of overhead off processors is another question.
Haven't noticed anything in the few cards descriptions I checked.

Re:CPU utilization, pc to pc, and Power not speed

[Anonymous Coward]

Indeed. The way an AHCI-compatible SATA device works (with DMA descriptors) allows far larger transfers over SATA than the smaller transfers allowed by USB3. In fact, USB3 has a lot more overhead in processing for disk transfers (even assuming the device is using the new USB3 streams stuff with UAS).

Stability issues?

[synapse7]

I have an etron and a Renesas USB 3.0 controllers in two different PCs. Both disconnect on large transfers in the range of a few GBs. Sure the transfer rate is holding down 80MB/s but I found them to be completely unreliable. If I plug the 3.0 drive into a 2.0 port the same transfer is made without issue, likewise if I plug a 2.0 drive into the 3.0 port there is no issue. Anybody else experience stability issues with USB 3.0 or is just me.

Re:Stability issues?

[Anonymous Coward]

I would suggest trying a better quality cable as they are much more intolerant at higher data rates.

Re:Stability issues?

[David_Hart]

My old computer has a Renasas controller. I had the same problem where USB 2.0 devices would work fine with large file transfers, but USB 3.0 drives on the same USB 3,0 port would have transfer problems. I upgraded the firmware on the Renasas chip and verified that I had the latest driver with no change.

Re:Stability issues?

[Mattsson]

I had similar issues with a USB3 pcie card. I don't remember the chip-model but the problem was solved by switching to another card made by another OEM. This one had the same chip and used the same driver, so I assume either a faulty card or a faulty design was to blame.

Re:Stability issues?

[Anonymous Coward]

You have an interference problem. I had this same problem with a USB 2.0 printer, even with a high quality cable. The problem was that the cable was next to a VGA cable and several AC power cables. Since USB is not isolated (like ethernet) and the differential data wires are not twisted (like ethernet), it is very sensitive to interference.

TLDR: move the USB cable away from other cables.

Re:Stability issues?

[Anonymous Coward]

I've had this happen to me. What fixed it was using a different (shorter) cable.

Re:Stability issues?

[Quila]

Sure the transfer rate is holding down 80MB/s

80 MB/s? Firewire 800 from 2003 can do better than that.

Re:Stability issues?

[Anonymous Coward]

I have an etron and a Renesas USB 3.0 controllers in two different PCs. Both disconnect on large transfers in the range of a few GBs. Sure the transfer rate is holding down 80MB/s but I found them to be completely unreliable. If I plug the 3.0 drive into a 2.0 port the same transfer is made without issue, likewise if I plug a 2.0 drive into the 3.0 port there is no issue. Anybody else experience stability issues with USB 3.0 or is just me.

I've had the same issue when using my Unitek USB 3.0 dock with the stock 6ft cable. I'll try a shorter 3ft cable from monoprice when I get the chance... maybe put some ferrite cores around the ends too.

Re:Stability issues?

[Anonymous Coward]

But his drive can't. Compare to this chart
http://9to5mac.files.wordpress.com/2012/06/usb-3-0-macbook-pro.png?w=390&h=242

Re:Stability issues?

[nabsltd]

Anybody else experience stability issues with USB 3.0 or is just me.

I only had an issue when I plugged two USB 3.0 devices into the same controller. When they were plugged into separate "root hubs" (use "View devices by connection" in Windows device manager), they worked fine.

Basically, I was trying to copy files from one external drive to another, and it kept dying just as you experienced. Now, I get close to 100MB/sec transfer speeds.

Boom!

[Anonymous Coward]

Another reason to upgrade next year!

usb?? ssd's on the pci-e bus are faster then sata

[Joe_Dragon]

usb?? ssd's on the pci-e bus are faster then sata

Thunderbolt killer

[sjbe]

So USB could replace SATA.

More likely it will just keep Thunderbolt from ever really taking off. SATA is pretty common and there are enough technical headaches with using USB instead that it is probably going to stick around. (though eSATA might be a different story since it is far less commonly used) But if USB is fast enough there really is limited need for Thunderbolt. I already can run a monitor via USB 2.0 through a docking station I use daily and that works fine.

I'm less interested in faster USB than I am in 100W USB. The ability to power a laptop or small PC with a single USB cable would be huge. Anything that reduces the number of different types of cables I have to deal with is a good thing.

Re:Thunderbolt killer

[skids]

They just need to get the laptops down to 25.5W, then they could run off PoE+, and they'd be able to put decent battery life in them.

Re:Thunderbolt killer

[Vegemeister]

Laptops are already under 25.5W, as long as you stay away from oversized screens and discrete GPUs.

Re:Thunderbolt killer

[QQBoss]

As soon as you can pump that 25.5W over 802.11N (PoWE+) without causing the cat's hair to stand up when it walks between me and the base station, sign me up!

Re:Thunderbolt killer

[sjbe]

They just need to get the laptops down to 25.5W, then they could run off PoE+, and they'd be able to put decent battery life in them.

USB is more flexible. I can (and do) run Ethernet over USB. Harder to do it the other way around. Conceptually you are right though - as long as we can do away with the single purpose power cord I'm fine with USB or ethernet or something else. A cable that carries power but no data seems pointless to me.

Re:Thunderbolt killer

[badkarmadayaccount]

The new PoE standard handles up to 45W. 802.3at, if I'm not mistaken.

USB 3.1

[tverbeek]

Actually applying version numbers to help people differentiate between subsequent verions would make too much sense.

Faster than my Ethernet - connection

[Gaygirlie]

Funny thing is that all the consumer-oriented networking equipment is still only up to 1Gbps and anything higher simply costs way, way too much. This makes me wonder if the new USB3 could be used as a replacement? How long can the cables be and still maintain 10Gbps speeds? And could one just connect two computers via USB3 without any additional equipment required in-between? Will someone come up with some USB3-based network routing solution before 10Gbps ethernet - solution become cheap enough for general consumer use? I would have use for higher speeds as 1Gbps just ain't good enough.

Re:Faster than my Ethernet - connection

[Charliemopps]

This makes me wonder if the new USB3 could be used as a replacement?

No
 

How long can the cables be and still maintain 10Gbps speeds?

About 10feet. Unless you had a powered cable or repeater... then about 20 feet.
 

And could one just connect two computers via USB3 without any additional equipment required in-between?

No, as stated above, you'd need a repeater.
 

Will someone come up with some USB3-based network routing solution before 10Gbps ethernet - solution become cheap enough for general consumer use?

There already is. You'll have to look around but there are such things. It'd be much easier however, to load balance multiple connections or switch to fiber.

 

I would have use for higher speeds as 1Gbps just ain't good enough.

That's because you're doing it wrong. Unless you're processing data from your basement super collider, there's no way you need faster than 1Gig. Most likely you have your network setup improperly and are NOT getting 1gig per second. Does your switch support Jumbo frames? Are they turned on? What are you transgering? What speed are your NICs? What speed is your buss? What speed are your hard drives.

The most likely problem that would cause transferring of files from one computer to another over a network is the hard drives. Their transfer rates are no where near 1gig per second. Your buss likely can't support that speed either. Last thing I'd check is your jumbo frames setting.

Re:Faster than my Ethernet - connection

[Gaygirlie]

That's because you're doing it wrong. Unless you're processing data from your basement super collider, there's no way you need faster than 1Gig.

That's where YOU are wrong.

Most likely you have your network setup improperly and are NOT getting 1gig per second.

I actually do get 1Gbps speeds.

Does your switch support Jumbo frames? Are they turned on? What are you transgering? What speed are your NICs? What speed is your buss? What speed are your hard drives.

Yes. Yes. Files. 1Gbps. PCI-E x2. Irrelevant, they're in a RAID and can perfectly well saturate the network as-is.

The most likely problem that would cause transferring of files from one computer to another over a network is the hard drives.

That would be true if they weren't in a RAID.

Their transfer rates are no where near 1gig per second.

Cache reads/writes well exceed the 1Gbps, and I get around 400 megabytes/second read-speeds from the array which translates to 3.2Gbps -- well over the network limit.

Your buss likely can't support that speed either.

You might wanna read up on PCI-E.

Last thing I'd check is your jumbo frames setting.

Already said that it is on.

Re:Faster than my Ethernet - connection

[nabsltd]

Last thing I'd check is your jumbo frames setting.

Already said that it is on.

And it probably only buys you 5-10% more speed, anyway, assuming you are getting decent overall speeds.

If you are down less 20MB/sec (like these benchmarks), you can get a 20-50% boost. Without jumbo frames, I run a solid 40MB/sec copying from one Windows share to another on my network, and jumbo frames couldn't get me over 50MB/sec, while mis-matches between jumbo/non-jumbo config dropped me to less than 20MB/sec at times.

Note that if I eliminate the slowest hard drive in the transfer by reading from a very fast 8-drive RAID array across the network to RAM in another machine, the speeds jump to nearly 80MB/sec. Protocal makes a huge difference, as using ftp (which has less overhead than SMB) I get 95MB/sec (from a 4-disk array to the fast 8-drive array). Yes, that's still below the 125MB/sec theoretical max, but I'm still reading from spinning drives and this not a test environment where the machines are doing nothing but a benchmark test.

Re:Faster than my Ethernet - connection

[Anonymous Coward]

Ae you running an old laptop? Hard drives have exceeded 1 Gbps for a few years. Not to mention SSDs have exceeded 6 Gbps. It will only be a years or two until SSDs start being installed in NASs. We need cheap 10 GbE.

Re:Faster than my Ethernet - connection

[MiG82au]

I love waiting around to transfer hundreds of GB large FEA results files or partition images etc. What kind of non super collider owning freak would need more speed?
P.S. A single consumer Hitachi 7K1000D 1TB will read sequentially at 180 MB/s near the start of the drive, and most 7200 RPM drives will do 140. My two examples are sequential BTW.
P.P.S. it's not the year 2000 any more, get with the times.

enhanced USB connectors

[AndyKron]

"enhanced USB connectors". I can't wait to see Best Buy's new $100 cables!

wait

[Charliemopps]

Are there ANY devices out there that would need even 1gig transfer rates? Much less 10? I'm sure there's some obscure shit out there that might, but is it worth a new standard? The only thing I can think of is maybe a RAM drive or SSD... but Sata would be a far superior choice for those devices. So again, why do we keep getting faster USB ports when there's nothing to plug into them?

Re:wait

[petermgreen]

eSATA has a couple of practical problems. Firstly if you set the SATA mode to "ATA" for ease of installing an older OS you often lose your eSATA port. Secondly most computers with eSATA only had one port and practically speaking eSATA is limited to one device per port. Port multipliers exist in theory but not all controllers support them, you can only have one level of them and i've only ever seen them integrated in drive enclosures not as a seperate product. Thirdly power was an afterthought hacked in later (eSATAp) so most eSATA drives needed a seperate cable for power. Finally eSATA has no backwards compatibility with any previous interface standard, some drives do support both eSATA and USB but you still need different cables.

In summary eSATA is fine if you have run out of drive bays and want an extra drive or are concerned about security and want to lock your drive in your safe when you go away so you have a drive that is external but lives with a single machine. It's not a good choice for a drive that is carried around and used with many machines.

10 Gbps is probablly overkill for current storage needs but SSDs keep getting quicker. Higher speeds also raise the possibility of graphics over USB3. The recent GPU switching stuff has shown that PCs now have enough internal bandwidth and the GPU vendors can be pushed to cooperate enough that a direct path from GPU to display output is not actually needed.

Re:wait

[Anonymous Coward]

I have a USB 2 docking station for my notebooks... It's great, but on a full HD screen it ghosts and gaming is certainly not an option. A USB 3.0 docking station would solve that.

Re:wait

[Anonymous Coward]

LOL - I think I remember making the same argument in the 70's... I can only type a couple hundred baud at most and nothing out there uses much more speed than that...why would we ever need to go faster?!?

LudicrousSpeed

[Anonymous Coward]

SuperDuperSpeed USB?

How about RidiculousSpeed or LudicrousSpeed?

http://www.youtube.com/watch?v=ygE01sOhzz0
http://en.wikipedia.org/wiki/Spaceballs

Re:usb?? ssd's on the pci-e bus are faster then sa

[Anonymous Coward]

It must be awesome to have to open your PC and pull out a PCI-e card every time you need to take some files with you.

To many it won't matter at all

[Anonymous Coward]

Given that Windows 7 and below don't support USB 3.

Re:To many it won't matter at all

[MachineShedFred]

Except that Windows 7 does. See: any Ivy Bridge-based computer that shipped in 2012 previous to the release of Windows 8.

Re:To many it won't matter at all

[QQBoss]

I have a Sandy Bridge-based notebook that did USB 3.0 just fine under Windows 7. I upgraded it to Windows 8, no more USB 3.0 (the port doesn't work for USB 2.0, either). Apparently the Intel USB 3.0 eXtensible Host Controller is still not qualified for Windows 8. There are hack workarounds that fake the installer into thinking it is still running under Windows 7, but it keeps getting disabled periodically and I have to re-install the drivers to make it work again (PITA, multiple reboots).

Re:usb?? ssd's on the pci-e bus are faster then sa

[snemarch]

So, are there PCIe bus implementations (and SSDs) that support hotplug for consumer systems? And are at a reasonable pricetag? :-)

Re:usb?? ssd's on the pci-e bus are faster then sa

[Anonymous Coward]

Expresscard... Thunderbolt... ePCIe...

Stupid plug design

[Immerman]

It's not that, USB plugs actually have a half-integer spin so that a 360-degree rotation actually inverts their perspective on the universe rather than returning them to the original orientation.

USB is actually a pretty unique case in user-facing plugs, and it rather pisses me off - what idiot thought that a perfect rectangle was a good shape for a connector? It's not like it was a surprise problem - almost every prior external plug was either a trapezoid whose orientation could to told with a glance or touch, or a round DIN which could often be partially inserted and then rotated until the "key" engaged (I rather liked those). Internal ribbon cables had already faced the problem for decades and come up with a progressive variety of solutions that most everyone agreed were sub-par and acceptable only because it was so rarely an issue. The keyed plug and collar had been settled on as the best solution for IDE and floppy cables long before USB was designed, and even that had the advantage over USB that you could often feel the respective orientations when working in situations where you couldn't see one or both of them. And it's not exactly like the solution was in any way difficult, just change the shape of the sheath. For crying out loud even the USB-B connectors designed as part of the same standard were uniquely orientable! /rant

main HDD not one to take files with you and

[Joe_Dragon]

main HDD not one to take files with you and having main hdd + a disk to take with you will just slow down the USB over all as well the likely X1 pci-e link

Re:main HDD not one to take files with you and

[Anonymous Coward]

My point is a storage device sitting on the PCI-e bus is no replacement for USB, even if the latter is much slower, because USB storage is meant for quick removal. The two really can't be compared.

DMA?

[bussdriver]

I thought USB was CPU bound did 3 add DMA abilities and all the fun security issues that involves?

Re:usb?? ssd's on the pci-e bus are faster then sa

[Anonymous Coward]

ExpressCard adoption has been a failure and the port is actually a lot slower than USB 3.0. Thunderbolt and ePCIe adapters are both very expensive.

pkzip

[Anonymous Coward]

2.04g

Upside-down connectors

[Anonymous Coward]

Can we get one that doesn't matter how it plugs in?

I swear, the connector on every USB device I have is upside-down. I always have to twist the cable to get it to plug in. Hell, I can even flip the device upside-down, and somehow, I still have to twist the cable to plug it in!!