Slashdot Mirror
USB2 Specs Are In
PooF writes "USB 2.0 has been announced. It seems sweet, its faster than firewire at 480 Mb/s.
An article from pc.ign.com on USB 2.0. Oh yea these are only the specs they hope to achieve no word on progress in implementing them."
This was covered in an earlier slashdot thread this summer. Faster USB is great. 12 MBits/sec is fast by serial port standards, but not fast enough for some of the things people want to use USB for (low-end video transfer, external mass storage, Ethernet, etc.). Despite that, there's a great mass market for Firewire, too. Here's why:
1: USB already has low-speed, misbehaved, "legacy" devices that need to be backwards compatible with the new spec. Firewire's legacy peripherals operate at 200 Mbits/sec. 'nuff said.
2: USB requires processor arbitration to run the bus. Firewire doesn't.
3: USB's design is specifically as a low-cost interface for PC peripherals. The hub-based design is a byproduct of this. Firewire is designed as a more general-purpose, device/device interface. Firewire can nicely connect consumer products to one another, no PC required.
However, USB has a higher "theoretical" maximum number of devices supported per controller, 127 (in USB 1.1), versus Firewire's 64. In practice, 64 Firewire devices is do-able, if silly. More than 4-5 USB devices (with a powered external hub) is pushing the limits. The only place where USB reaches the upper limits is at USB technology bake-offs.
I love USB (heck, it's in all my PC's and both my home Macs), and it's a great cross-platform standard for computer-oriented low-end to midrange peripherals. Firewire is better for high-end devices (prepress scanners, hard drives, video equipment, etc.), but it's a general-purpose interface, and that's why it will ultimately do well. The two interfaces are not, by any stretch of the imagination, mutually exclusive. Anybody who thinks that you can only use one either needs to buy a Mac to prove otherwise or board the cluetrain.
- -Josh Turiel
-- Josh Turiel
"2. Do not eat iPod Shuffle."
OK, I'm actually at the USB 2.0 Devcon right now, and have read the 2.0 proposed spec, so I'll take a stab at refuting these comments:
1.The usb.org article only claims "120-240Mbps". It's not clear where the ign.com article came up with 480Mbps.
The speed is 480Mbs. That is what the spec says.
2.Even if USB2 runs at 480Mbps, the Firewire folks aren't exactly standing still. Any raw bandwidth advantage of USB2 is sure to be short-lived at best.
Firewire and USB have too many things that are not in common, they really are not competitors. USB is aimed to be a PC centric bus. There has to be only one host, and a whole lot of clients. Firewire can be host to host. Firewire is more intrenched in the consumer electronic market, while USB is sticking to the PC (for now).
3.There's lots of blather in the USB2 announcement about supporting video cameras etc. but IIRC USB doesn't support the isochronous transfers which are usually considered necessary to serve those markets. Did I miss something?
You missed something. USB has always supported isochronous transfers. Look at the USB speakers from Philips for an example of a shipping product that uses this. Isochronous is still there for 2.0.
4.Another useful Firewire feature that USB doesn't seem to have is providing power through the same connector used for communications. Again, I may have missed it.
USB has ALWAYS supported power on the connector. How else does some of the devices work? 2.0 does not change this. It's still 5V at 100mA-500mA depending on what you need and ask for. If you need more power, take a look at the Plus Power Connector that IBM supports for USB. It can provide 12V or 24V at 3A. That's about all the current that anyone needs.
5.I don't remember how many devices USB supports, but I suspect it's less than Firewire.
USB supports 127 devices per host controller. You can plug in more than one host controller in your PC at a time. The record (I think) for most devices plugged in and working at once is around 144.
6.I know that USB-based host-to-host networking exists, but it's not clear to me whether it's really as well suited to that task as Firewire. In particular, I wonder how much asymmetry between hosts and devices (a la initiators and targets in SCSI) is built into the protocol, and how round-trip latency compares to other technologies.
As I said above, USB is a host-client bus. You can make (and buy) a device that does networking over USB from one computer to another, but this is just two client devices talking together in a box. Firewire can do true host to host on the bus itself. The USB protocol is a star topology with the PC host controller at the top. I can look up the round trip latency stuff somewhere, but it is built into the protocol, and the host and hub controllers seem to handle it well.
7.Similarly, I'd like seeing a comparison of how automagically reconfiguration happens when devices are added or removed using each technology.
I don't really know how Firewire does this at all, but USB handles this wonderfully. There is a description of how the protocol handles all of this in the spec (at www.usb.org).
In summary, USB 2.0 looks like it handles a lot of the speed issues that some people had with 1.1. It provides backward compatibility with all 1.1 and 1.0 devices and enables things like speakers and video cameras to run better.
Like it or not, USB looks to be here for a while. A lot of computers are coming out without a lot of different connectors, and USB is replacing them.
Ob Linux: USB is working on Linux in the 2.3.x series of kernels (it's also supported a little in the 2.2.x series, but not for many devices.) More information is at www.linux-usb.org
I think not. USB 2.0 may be faster in terms of throughput than Firewire, but not on CPU utilization. USB sucks up more processor, and thereby slowing your entire machine down. This is another example of Chipzilla driving a "need" for faster processors. Remember about a year ago when DVD technology came to the PC arena. Some video card companies came out with hardware DVD decoding solutions. Intel made the push that their Pentium II processors could easily handle software decoding of DVD, and therefore the hardware DVD solutions were not needed. I'm all for the fastest processor, but not when it's need to mask performance bottlenecks in other areas. Who wants to do one thing on their PC? Why do we even try to multitask? And I guess the fact that Firewire was invented by Apple has nothing to do with Intel pushing their new "faster" solution.
Actually, this is specsmanship. Where IEEE1394 does its arbitration, framing, etc. at a minimum of 100 Mb/s and usually at 400 Mb/s, USB of any flavor does its polling (notice, it is *not* a peer interconnect) and framing at no more than 12 Mb/s (assuming that there are no low-speed 1.5 Mb/s devices in the path).
Each packet in USB2 starts out with the same preamble (12Mb/s) and ends with the same postamble (12Mb/s); the new twist is that there may be a tone burst in the middle of 480 Mb/s or whatever. Unless you ship really huge packets (which aren't allowed by the protocol anyway) the average transfer rate is a lot lower than 408 Mb/s.
Lacking <sarcasm> tags,
With any luck, someone more clueful can fill in the blanks above.
Overall, Firewire still looks like a generally superior technology in its niche, while the USB folks should have been content with their own separate niche (lower-bandwidth peripherals such as keyboards, mice, joysticks and modems that don't need advanced features such as isochronous transfers).
BTW, I don't actually use Firewire and have no interest beyond the aesthetic in promoting it. My employer is thoroughly committed to (ick) FC.
Slashdot - News for Herds. Stuff that Splatters.
Ican help you out there - it *is* a move from Intel.
1 0/idf-99-01.html
The difference between firewire and USB is sorta like the diffence between the SCSI and IDE interfaces for harddisk etc.
Silvino Orozco from Toms hardware gives a nice explanation:
"The big drawback is that USB actually uses CPU horsepower and some people are just not happy with that. This is why people push IEEE 1394 (Firewire) so much. It has high bandwidth but doesn't eat up the processing power that USB does."
Here is the link:
http://www.tomshardware.com/editorial/99q3/9909
Intel is of course not happy to see more things being moved away from the CPU (NVIDIA's new GeFORCE256 GPU is another example) - Intel pushes everything that demands a lot of CPU-power, and I bet that pushing that much data over a connection as USB2 is supporting, is going to use a lot of CPU-cycles...
That's why I hope that firewire will catch on - maybe the Playstation2 (which AFAIK uses firewire) will give firewire a nice boost - I hope it does...
Current IEEE 1394 developer and maintainer, at your service :-)
It works with asynchronous transfers. I'm currently working on updating the userspace raw1394/libraw interface. When I'm done I'll try to get it into Linux 2.3 again (even though it's in feature freeze, but this subsystem does not affect any other code in the kernel).
Isochronous transmissions are not yet supported and I know that this is important for the people who want to get pictures from their cameras. At least isochronous receiving should not be too hard to implement, that is around next on my todo list.
As for the supported hardware: AIC5800 (out of production AFAIK), PCILynx (hard to come by, obsolete), OHCI (the standard of the future, implemented in hardware by various chip manufacturers). A Sony chip also exists (e.g. in the Vaio laptops), but is not yet supported. My OHCI card (donated by ADS Technologies) for some reason doesn't work, but I'm concentrating on PCILynx for the time being.
The USB 2.0 specs are indeed respectable. Think of all the irq's you wil save. But it won't eclipse firewire anytime soon.
Right now 400 Mbps is not a bottle neck for most consumer end hard drive setups. Firewire and Sony's iLink (both IEE1394) are being pushed in the DV realm, where this fat a pipe is really needed.
On an earlier post, someone speculated that Firewire supports more devices than USB. It does not. The USB 1.0 supports up to 128 devices (which, ironically, was proven true by Apple's employee's during in a bier-garten, not by Intel employees), whereas Firewire supports 63 in it's current incarnation. But if USB 2.0 does support it's 480 Mbps claimed speed, one doubts that it will be able to support 128 devices.
There are Firewire port prototypes at 800 Mbps (this is due out next year), and the 1600 Mbps versions are in the works. In any case, more development has been done on 1600 Mbps Firewire than USB 2.0.
Everything seems to point to Intel launching a FUD war against IEEE 1394 technology. I suppose announcing a product months ahead(or perhaps years in this case) is typical of any large corporation, but there are other indices. 480Mbps is a minimal improvement over Firewire 1.0's 400 Mbps, but just enough to convince consumers and vendors. And, as speculated by SuperScan, USB 2.0, like USB will utilize CPU power to get by, whereas Firewire delegates this task to the Firewire controller.
Firewire is a part of the "PC2000" standard that was proposed by Intel and Microsoft. Maybe Intel wants to revise that proposal.
Recall that Firewire is not solely an Apple technology. Sony is hard core on IEEE 1394 (PSX2 will have Firewire ports), as well as others. If intel wants to dominate all bands of the peripheral device spectrum with USB + USB 2.0, they're going to meet some stiff resistance.
This kind of move just shows how intent Intel is on being the MS of the hardware world. By selling a CPU reliant standard, they get to dip in your wallet twice.
Btw, does anyone know if it's true that Apple cancelled it's 1 dollar licensing charge for Firewire ports? (By (un)popular demand?)