Slashdot Mirror
USB 2.0 Spec Is Final - Up To 480 MB/s
mrbinary writes "I hadn't checked the USB website for awhile, but it seems that the USB 2.0 spec is final and the speeds are confirmed at up to 480 MB/sec. Not too bad - hardware should be available shortly and the spec is fully reverse-compatible with USB 1.1. There's a PDF document that talks about the goodies upcoming including: 'A first-ever demonstration of multiple MPEG videos being played on a ZIP drive connected to a PC via a USB 2.0 Bridge. The demonstration featured an FPGA bridge designed and manufactured by In-System Design, an I/O design and engineering firm specializing in USB bridging solutions.'" USB is actually a cool standard, but cooler on paper sometimes than in real life. Hopefully 2.0 will squeeze ideal and reality a little closer together. [Updated 4:00 GMT 22 Aug by timothy:] As reader David Bastiani points out, 480 Mb/sec sounds a lot more reasonable than 480 MB/sec. Maybe that will make it easier to make the specs meet reality;)
Intel just wants to keep USB alive to try and steal the power of 1394. But in reality, the main benefit of 1394 is not being tied to a PC host or having a hub based network. Where is this useful? In the home entertainment area, where you could hook your 1394 HDTV to a 1394 DVD player and enjoy. One cable per device, instead of the several to get video and audio. Plus it's all digital.
USB2 will be nice for the speed, but 1394 is prepairing to hop to 800Mb/s then on to 1.6Gb/s.
Uh...480 Megabytes per second? I guess we can scrap all these fibrechannel arrays right?
Obviously you're not familiar with Last Kiss or Teen Angel or that song about the kid who entered a race trying to win enough money to buy his gal a ring but got killed instead.
Classics, all of 'em. One listen and you'll be straining through your tears to see the monitor as you write your retraction.
USB wasn't meant at all to originally be used for those kinds of things. It was meant to be a simple replacement for old PS/2 and serial ports. It was good that USB was kept lightweight. It made putting USB controllers on machines a lot cheaper.
I don't see what advantages you tout about USB aren't possessed by Firewire. Firewire devices are hotswapable, and are connected in equally shared long chains -- and Firewire doesn't have the hideous resource sharing issues of USB and USB2.
You see, USB and USB2 are both processor-arbitrated bus schemes. This is yet another insidious move by Intel to keep people dependent on buying faster and faster CPUs, whereas Firewire allows the devices to negotiate intelligently for bandwidth usage. Furthermore, USB2 requires far more intelligent balancing of devices on the chain than before -- eliminating one of the primary USB advantages, simplicity.
The problem is that all USB/USB2 devices get an equal chunk of time. This means that if you have a keyboard, a mouse, a scanner, and a digital camera on the same chain, each one gets 1/4 of the 480 Mb/sec bandwidth. Your keyboard may only need the low-end 1.5 Mb/sec rate, but it gets a full 120 Mb/sec. This means that you have to learn to keep your high-end devices off of the same chain as your low-end devices, or you have to get used to unplugging things when a specific device needs the bandwidth. This also ignores the complexities of USB/USB2's star topology.
With Firewire, it's plug-and-play -- at the expense of more complex driver and hardware controllers. Firewire can even guarantee a dedicated minimum amount of bandwidth. USB2 is an insidious marketing ploy by Intel to destroy Firewire. Since Intel is making it, ignoring the fact that it'll be years since its initial announcment before we see products while Firewire is here and now, it's "inevitable" that it'll replace Firewire. This is a blatant powerplay at the expense of consumers to attempt to make people more reliant on faster Intel chips. It should not be celebrated for "Finally" getting closer to achieving its goal. I hope USB2 dies a flaming death.
If it's for-profit but free, you're not the customer -- you're the product (e.g., the Slashdot Beta's "audience").
Take a look at this little article on USB 2.0 and this one.
I can say that firewire is here and it works now, beautifully. Why should I wait for manufacturers to develop and implement USB 2.0 mobos and devices? For a measly extra 80Mb/s? I'll wait for firewire to go to 800Mb/s later this year.
What? No firewire devices besides camcorders? My favorite firewire devices are the sancube and this portable firewire raid array.
Now they (hard drive manufacturers in particular) just need to make some native firewire devices, bridges are just so... inelegant.
USB 2 is a poorly executed hack perpetrated by Intel purely because they couldn't handle having to pay the FireWire licensing fees.
FireWire (even the 400 Mbps 1394a version that is currently shipping) is superior to USB 2. It has several important advantages, perhaps the most important being that it is root-independent.
FireWire explicitly includes peer-to-peer transfer. You do not need a computer to hook FireWire devices together and let them communicate. On the other hand, USB is explicitly designed as a computer peripheral bus--the root node must always be a computer. Without a computer powered on and managing the Universal Serial Bus, it's useless.
FireWire's isochronous support came first and is more robust than USB's. Up to 90% of each USB frame is reserved for isochronous data (combined with interrupt packets like mouse clicks), compared to 80% for FireWire. However, USB's isochronous cycles trigger every millisecond, where FireWire's trigger eight times per millisecond, making more efficient use of bus time by dividing it into smaller intervals. FireWire's higher bus speeds make it a far more robust isochronous transport.
FireWire allows bridging buses together, up to 1023 of them. There are already USB bridges in discussion (perhaps available), but they skirt the edges of the specification. If you can plug a single device into two computers through USB, each computer will see the peripherals downstream from the "bridge," but the bridge has to include lots of special logic so the two computers don't try to use the same peripherals at the same time. I'd be awfully wary of plugging a disk drive into such a setup.
It's hard to say whether USB 2.0 will overcome these deficiencies--but if it does, it will basically be FireWire. Intel was a great proponent of FireWire until just about the time the company realized they were going to have to pay royalties on chips like every-one else. That's when Intel decided that FireWire was useful only for consumer electronics (like camcorders), but the quickly-invented USB 2.0 would be the preferred way to connect printers, scanners, disk drives, and so on. Since the licensing fee matter was decided, all those companies that had been ambivalent on FireWire are again behind it--including Microsoft--except Intel.
He's no good to me dead.
My goal is to have each device on my machine use a different interface!!! (SARCASM ON)That would be Cool(SARCASM OFF)
SCSI for my Jaz
FireWire for my Camcorder
USB for my Rio
Parallel for my Printer
PS/2 for my mouse
Serial for my Dig Cam
and finally....drumroll please....
USB2 for that Scanner I have needed for a long time...
(What a wasteland hardware has become!!!)
(+1 Funny) only if I laugh out loud.
My mouse will no longer be my bottleneck, yes.
Does anyone know what became of this? Does the final USB2 spec contain any such anti-consumer features?
In USB 1.x a low speed device (1.5mbps) occupies the entire bus and takes time from standard speed (12mbps) devices.
In USB 2.0, though. the hubs have buffers and do active rate matching. A standard speed device will NOT bring down the speed of the bus and will not get in the way of high speed (480mbps) transfers. The silicon space to do this is quite negligible these days.
Intel has supported 1394 for years - it was going to be the successor to IDE. In USB conferences prior to 1998 they kept talking about how USB and 1394 are complementary - 1394 will never be cheap enough for a mouse. If something made them change their mind I believe they have a real reason.
----
Stop worrying about the risks of nuclear power and start worrying about the risks of not using nuclear power.
Firewire (IEEE-1394) is here today, works better at high speeds, an IEEE standard, with working silicon and device drivers. Why not use it?
Mea navis aericumbens anguillis abundat
480 Mbits for USB, or 60MBytes per second. Firewire is supposed to go to 3200Mb sometime in the not-so-far future. That's 400MBytes per second. A great idea, you think, for that personal Beowulf cluster you've got sitting around in your basement - just implement TCP/IP over firewire, and you're done. But wait! 32-bit PCI busses allow for only 132MB/sec (32 bits at 133Mhz) or 264MB/s (if you're lucky enough to get a 64-bit extension on that bus of yours).
Time for a new architecture, I think.