Now that USB 3.0 is out, when will Linux get it?
Well, Linux has it. Windows does not. Mac does not.
Are you shocked? If so, you have just exposed a limitation on your own thinking. Linux supports more hardware (overall configuration and bits and piece) than any other system, by far. Linux is quicker to support hardware other than cases where proprietary drivers come out with the hardware from closed source companies, but Linux then ultimately tends to support those drivers sooner than other non-targeted OS systems do.
In fact, let me tell you just how bad your thinking was on this, if you were at all surprised that Linux has USB 3.0 support (and has had it since last September) while the other operating systems do not. You probably think that some hardware only works with some systems. For example, a Mac G3 can run a Mac system, and a "PC" by Hewlett Packard can run only a Microsoft Windows system. So far so good. You may even think that there is some magical reason why this is the case, something fundamentally Windowey about PC's and something fundamentally Macey about Macs.
But no. There is no pragmatic reason that any given computer operating system (among those being discussed) can't operate any piece of hardware that any other system operates. None.
And Linux developers know this so they just go ahead and develop the software to run the hardware. The failure of proprietary systems to serve the diversity of hardware that is out there has turned into a fetish system whereby we believe things that are not true, not good for us to believe, but that benefit the proprietary companies.
Anyway, details of USB 3, Linux and also Firewire are here.
- Log in to post comments
More like this
I need to have more faith in you, it seems. When I read the line immediately above the fold, my first split-second reaction was that I was SO gonna click on "read more" then give you a piece of my mind because Linux already supports it. As I was clicking another part of my brain kicked in (the Admiral Ackbar part) screaming "It's a trap!"
Meaning "in real life" fantasy in which one has a device which can either deliver or soak up 350 MBps. In that other real life, in which we actually live, throughput will be limited by the performance limits of actual devices.
It doesn't seem particularly useful to support a device type that is only available as demo boards, and was only released as a real standard a couple weeks ago. Examples of worldwide stupidity on this level: I have no idea which version of "Draft-N" my various wireless cards are running, but they mostly seem to talk to each other. Sometimes. Early ones are wall decorations at this point.
That said you can bet on the other OSs having support available by the time motherboards are released with USB 3.0 on them.
And about "general hardware support," that has a few caveats. x86 is well supported, but I can tell you from experience in attempting to support SPARC processors that sun's older ISA is "not well supported." But it is possible to compile a linux kernel on a SPARC V8 machine, just dodgy. It could probably be worked out with enough effort. Likewise there is no bloody way in hell that either a mac OS or windows OS is ever going to even attempt to run on them. Same issues on my little blue toaster.
It doesn't seem particularly useful to support a device type that is only available as demo boards, and was only released as a real standard a couple weeks ago.
This is exactly why the open source model is superior to the proprietary model in almost all cases.
you can bet on the other OSs having support available
Exactly. You can bet. I wont be betting, I've got a sure thing!
I can tell you from experience in attempting to support SPARC processors that sun's older ISA is "not well supported."
Right, Microsoft and Apple have you covered there ... Not!
Same issues on my little blue toaster.
Alomst ready: http://linux.about.com/b/2004/08/27/linux-in-your-toaster.htm
My very next sentence after you quoted was "Likewise there is no bloody way in hell that either a mac OS or windows OS is ever going to even attempt to run on them."
Little blue toaster:
http://popcorn.cx/computers/sgi/o2/o2-02.jpg
MIPS processor. BSDs supported, Debian "functional" but missing many things.
As for embedded things, you'll find that for a good long while most televisions, VCRs (What's the?!), and other x86 powered household devices ran on stripped down versions of DOS. Linux is certainly the future of these applications.
Excuse, not very next sentence, but a couple sentences later. I should read my own posts.
Spiv, I know, I was just giving you a hard time.
Linux is by far used on more different embedded devices than any other operating system, and the second most used is not a Windows product.
Can you imagine trying to produce a dishwasher that rand Windows and then getting the company lawyers to allow that? The liability costs would be out of this world.
There's worse out there. For example, my bank's ATMs run Java on some Windows (XP or later CE, there's that wavy Windows logo on the Start button). Of course, that's only visible when the JavaVM crashes. Which happened to me more than once.
From http://hardware.slashdot.org/hardware/08/05/08/1247238.shtml
There are 3 ways to make things "Just Work":
1.(The Apple way) Be able to control everything, from the metal to the
display. If you can make everything the way you want it to be, you can
make things work by design.
2. (The Microsoft way) Be able to control everyone, from the hardware
manufacturers to the software developers. If you can make everybody make
things the way you want them to be, you can make things work by fiat.
3. (The F/OSS way) Be able to know everything, from the hardware registers
to the software code. If you know everything about the components you use,
you can make things work by hacking.
The reason Richard M. Stallman is so adamant about making things free is
because we, as end users, have no other way to make them work for us.
I admit that I was surprised to find that Linux supports USB 3.0, but that's mostly because this is the first I've heard of USB 3.0.
Who cares about USB 3.0 support at this point if there are no USB 3.0 devices? I am sure once 3.0 devices hit the market the proprietary OS's will have support for them.
Linux being first-to-market with 3.0 is nothing but a worthless bragging point.
Besides Light Peak is going to make USB 3.0 pointless.
Roland, nicely put.
Michael, there is a reason that this was pointed out.
Wooosshhhh...
That was you missing it.
When the technical panels released USB3.0 (this was a few months ago now) the reference implementation was on Linux, so Linux was the only OS with USB3.0 capability. It makes sense because the source code is free, so anyone can have a look at the details of an implementation.
Well, yes, and to put a slightly different spin on it, Microsoft is probably in the same place as Linux, as is Apple if they decide to support 3.0 agt all, but for the Linux implementation we know about it, can use it now, can see it, could have contributed to it if we wanted to, and don't have to worry about the usual rollout bullshit (like getting it wrong the first few times but never admitting it, etc.).
@Virgil Samms: The engineers involved push the maximum throughput to some limit they believe is achievable. USB3.0 must also support v2.0,1.1,1.0 (unless the USB consortium decides to break its promise). The limits are not set based on any existing device and the data throughput which any one person might imagine was necessary. In fact I have many applications which would do well with v3.0; the data will come in bursts, but when data is available I want it transferred at very high speed. Gigabit ethernet will work for me but I need to run a proprietary protocol (not IP) because IP has too much dead weight in the protocol (wastes transmission time). The USB protocol on the other hand happens to work just fine (except that no devices are available yet). The "practical throughput" is essentially the maximum predicted given the least overhead for each USB transaction.
O_O linux has it and windows dont lol. well about the hardware and functionality yes linux is a better platform
An interesting postdiction. I bought and installed four such motherboards last month. They are up and running with Linux. Do Windows support it yet?
For example: ASUS P7P55D-E Premium LGA1156 Intel P55 DDR3 - 2133 ATX Motherboard USB 3.0 SATA 6 Gb/s
USB 3.0 is available today. Presuming Intel goes full speed ahead with LightPeak, there might be products on the market in two years. Or longer. And if/when they are available, they will be supported on Linux.
Motherboards are also available from Gigabyte; they have models with USB3.0 for both Intel and AMD CPUs.
Devices to hook up to those motherboards: 1 TB external hard drive. I might wait a month or two longer though for some hot SSD drives which will actually exceed USB2.0 bandwidth. Such as the SuperTalent RaidDrive.
I'm pretty sure my toaster is already using LightPeak with Linux. In six languages so far.
Intel claims that we will see Light Peak over the next few months:
http://news.bbc.co.uk/2/hi/technology/8272003.stm
Speaking of dead weight, the USB protocols are not so great in that respect. For example, while the theoretical throughput for USB2.0 is ~ 48 MB/s, good luck betting better than ~ 35 MB/s. You might want to read up on the UASP protocol initiative. It should improve throughput for mass storage devices under USB 2.0 and 3.0.
I already knew that Linux supports USB 3.0
one more thing Linux supports hardware that doesn't have drivers for newer versions of Windows and Mac OS X
I'm late to the party, and pretty much nothing new here...But
Until Intel makes the steps to support USB 3.0, it will not reach Windows.
And I see this as your point, more or less - since Linux is in the open, and the standards are there, they have support before others. Apple's currently concerned with lightpeak (which is slowing Intel to finish up their work on 3.0)
And there is no reason for MS to jump out of the gate with USB 3.0 support, they'll need to take a look at the hardware, test test test before they can offer support. Linux has no need for that, as supporting the hardware itself is testing.
@Virgil Samms: Thanks, I'll look up that protocol. The USB protocol (polling specs, data/handshake specs) is a case of "one bus tries to do everything"; it's difficult to keep overhead low while maintaining general purpose use. However, unlike Ethernet, the USB protocol does not allow you to put a different protocol stack in place (and in fact much of the USB protocol is implemented in hardware).
USB 3.0 not on motherboards or market devices?
Wow, Microcenter, Asus, and Buffalo were all conspiring to lie to me then.
I checked out one of the USB 3.0 Asus motherboards. Reason not purchased: only 2 USB3 slots, ad I think 10 or 12 USB 2 slots.
And then the Buffalo external HDD with USB 3.0. Reason not purchased: drive speed only 7200rpm, meaning the hard drive rather than the USB would be the severe bottleneck.
As for Intel's LightPeak, I'm not sure we'll see full market penetration. It's not a backwards compatible device. It's a new interface. It may be like Firewire - great, but never fully penetrating the mainstream.
Also, all the PR aside, I don't think the wear-and-tear factor will be significantly higher for LightPeak than USB, and is far more likely to be significantly lower. Given the things I've heard from fiber techs regarding fiber, we'll see how things work. And I'd give it two or three manufacturing/release generations into EITHER hardware until I buy into it. Because early adopters pay the price, in doubles.
From your link:
That's a hope, not a firm commitment, and it gives them a full year.
Light Peak was first publicly demonstrated at the Intel Developer Forum in 2009. For comparison, USB 3.0, which is just now hitting the retail market, was introduced at the September 2007 IDF. To get Light Peak to that stage, Intel and allies will need to develop standards, both for software protocols and for the hardware; cables and connectors. Since external devices these days want a cabled-in power supply, Light Peak connectors and cables will be compound devices, with both optical fiber and copper conductors. That will figure into planning time and into wear-and-tear.