Intel's USB 3.0 and Wireless USB 1.1 target speeds announced: so long Firewire?

Intel's announced USB 3.0 specification could push throughput beyond 4Gbps (300MBps) at the application level while introducing Quality of Service in support of HD video streams. Besides supplanting Firewire once and for all, a clear goal of the new "SuperSpeed USB" is to keep up with the transfer speeds of flash chips. "We don't want to be the bottleneck in the system," says Intel's Jeff Ravencraft who is overseeing the 3.0 initiative. Intel, HP, Microsoft, NEC, NXP, and TI will present the initial spec for a design review in November with first silicon to be stamped in "early 2009." While the new interconnect (pictured) will remain backward compatible with USB 2.0 and prior devices, new cables laced with an optical link and a max length of 2-meters will be required to take advantage of those high speeds according to a senior engineering manager with NEC. Meanwhile, a 1Gbps throughput is being targeted with Ravencraft's other baby: Wireless USB 1.1. Sounds great, but with existing 480Mbps Wireless USB silicon only achieving about 40Mbps in practice, Intel would be wise to focus on efficiency, not theory. Of course, it's all just a lot of smack-talk 'til they deliver, but with Apple running Intel inside now, Sony putting USB in their camcorders, and eSATA proliferating for external disks... well, Firewire's days sure seem numbered.

[Via EETimes and The Inquirer]

[mattclarkie @ Sep 19th 2007 8:13AM]

acsirecording is that like keylogging. Get the joke, I am hillarious

[oshean @ Sep 19th 2007 8:22AM]

Sure mattclarkie, make fun of things you don't understand.
The interweb doesn't seem powerful enough... some please define: acsirecording.
Thx

[mattclarkie @ Sep 19th 2007 8:27AM]

If you read the whole sentence it should have said "according to a senior engineering...."

You Jackass

[James Ollier @ Sep 19th 2007 9:46AM]

USB may be faster, but firewire has a better name and a better shaped connecter, no upside-down issues with firewire...

[oshean @ Sep 19th 2007 9:57AM]

matclarkie. Obviously.... at the time it didn't read that way. Anyone who can't take a joke or has to proclaim that they made a joke is the Jackass.
Engadget....thanks for correcting the typo.

[Typhoid Mary @ Sep 19th 2007 8:58PM]

Forgive me James but I'll take the faster one. Feck the name and if you cant plug a wire in the right way then your an idiot.

[E71 @ Sep 22nd 2007 5:38PM]

Typhoid Mary, firewire has always been faster than USB 2.0 (despite having a lower declared throughput of 400Mbps vs USB's 480Mbps). That point alone is enough to say there's something wrong with USB... but then there have always been other issues with USB.

[citizen @ Sep 19th 2007 8:21AM]

theres one thing i wish the next usb standard would have, and thats the capability to supply more power, then maybe all those external 3.5" hardisks wouldnt need powerbicks!

[matthew @ Sep 19th 2007 8:42AM]

here! here!

[Joshua Miller @ Sep 19th 2007 9:32PM]

Actually, that's "Hear, hear!" as in "Hear what he is saying!"

[Temple @ Sep 19th 2007 9:45AM]

Amen, I've personally never had any issue with speed as much as power. For instance, even external DVD-burners for my laptop need additional power. I have a slim Plextor DVD-burner which needs to connect to a second USB slot just to get enough juice to run.

[dv @ Sep 19th 2007 10:55AM]

no! people should stop thinking of USB as a power supply.

i'm tired of seeing computer-less USB devices like USB shavers, USB fans, USB drink coolers, and USB projectile launchers. where the heck are you going to be such that you have a USB port but no power plug, and desperately need to shave?

increasing the power standards is only going to make it worse. we're going to start to have USB saws, USB drill presses, USB who-knows-what that probably should not be sending inductive loads through your computer. people will make this stuff, i tell you. but people need to start realizing that a USB hub is not a magic source of a bunch of power sockets. there is also this wonderful thing called a power strip that you can plug into any power socket, ubiquitous everywhere you go, and you'll get plenty of places to plug in your shavers and rice cookers and what have you.

[OddManOut @ Sep 19th 2007 12:21PM]

Agreed!

It's nice to see the standard evolving. USB has been one of the best things to happen to PCs in the last 15 years (no offense to IR/Parallel/Serial, which had their days in the sun). An extra 5-port usb i/o card is standard equipment on all the machines I build. It would be nice if you could simply connect one of those increasingly unused molex power connectors to your i/o card and supply the full 12 volts out to your devices.

Hey, we've got power over ethernet now...why not!

[HineyWipe @ Sep 19th 2007 8:21AM]

Oh, I am so looking forward to those Best Buy sold, 1 Meter USB3.0 gold-plated cables, for like a steal at $99. I can hear the Blue shirts squeal with glee at that sale!

And in other news, FireWire is not dead (IEEE1394, IEEE1394b and even IEEE1394c). It's too prevalent in DV, ... and for now, its throughput is better sustained than USB2.0. But hey, in Intel's theory, atleast they are keeping the same connector-backwards compat.

[Shinrai @ Sep 19th 2007 1:07PM]

"The blue shirts" don't make any commission. They really couldn't care less HOW much your cables cost.

[freakscene @ Sep 19th 2007 2:14PM]

Well, don't shop there then. http://monoprice.com sells the same quality cables, but for wholesale prices. I recently bought an HDMI cable there for 7 bucks. The same cable branded by Monster would have cost me 10x that.

[ekwmin @ Sep 19th 2007 1:59PM]

Yep, Firewire has it's place. It's not all about maximum speed. Many people don't realize that Firewire's performance is pretty much constant where USB tends to fluxtuate. From my experience I feel that Firewire is more solid and responsive. I always spend the extra dollar to get my external drives in IEEE1394b, while other devices I'm perfectly happy with USB 2.0. I'm sure both formats will co-exist.

[yoshi @ Sep 19th 2007 8:31AM]

The 1394b spec is already rated at 3.2Gbps. The components just have to catch up to the spec. As you noted with wireless USB, what's on paper and what's reality aren't necessarily the same thing.

[WhiteStar @ Sep 19th 2007 8:40AM]

Sounds good to me. Backward compatibillity is always good. Hope they're finally gonna burry the serial and PS2 ports for good when this comes out.

[I LOVE THE CAPS LOCK KEY @ Sep 19th 2007 11:23AM]

There are still millions of devices out there that need a serial port to operate. RS232 is still very prevalent in the medical and industrial sectors. Also there are still tons of dedicated vintage PS/2 keyboard buffs whom are petitioning manufacturers to keep the port alive. Hardware based mice are superior to their software based USB mice, better frame rates when the CPU is under load with less jerking.

[WhiteStar @ Sep 19th 2007 11:49AM]

Strong agruments, though I'm not sure if all of them are valid.
* Most/all? serial and PS hardware could still be used with an adapter. So much for vintage geeks.
* USB mice perform nicely under CPU load on modern systems, otherwise those hardcore gamers would use PS2 high end mice - which don't exist, because USB mice are better.
Every PS2 port that goes makes room for two USB ports, every serial for four. If you're talking mobile devices... that's something.

[Andrew @ Sep 19th 2007 1:48PM]

Actually some of old ports such as parallel port are quite useful to us who do EEPROM programming or any kind of DIY as they are FAR easier to interface with and require no special chips, licenses, patent rights or whatnot. Why should I need to buy new equipment if old one is working perfectly? Not to mention that not all applications require USB speeds - ease of use are sometimes more important and USB is anything but simple to use on the hardware level.

More importantly though, there's the reliability issue. The USB is seriously screwed up standard with unacceptable number of glitches this many years later. For example I have *NEVER* *EVER* been able in these 10 years to get any USB hub to consistently work as it should (i.e. transparent). Devices disconnect or their transfer rates stutter, machines freeze or won't boot, they are not recognized, they conflict with each other, anything you can imagine. It's not just hubs, I have had numerous compatibility problems with direct USB motherboard ports too - USB mice stutters if there's heavy traffic going on - doesn't matter if it's other USB device, network card or hard drive, there's always SOMETHING that doesn't quite work.

Yes, in the old days you'd have to deal with issues of interrupt distribution to each device so they don't conflict. But at least critical stuff like keyboard and mouse - with dedicated interrupts and ports - would always work.

[Doug @ Sep 19th 2007 3:25PM]

I seriously hope that USB 3.0 will eliminate some of the wonkiness that 1.0, 1.1, and 2.0 have all been so kind to deliver us. As mentioned above, there are serious issues when trying to use multiple USB peripherals simultaneously, especially if going through a USB hub. And PS2 mice worked MUCH better than USB mice ever have. Oh, and 500ma at 5V? That's only 2.5W... almost not enough to do anything.

The other serious problem with USB is how much load it places on the CPU of a system. Capturing even highly compressed video over USB can consume 25% or more of even a fast CPU. Firewire doesn't do that. This same issue is responsible for jerky mouse pointers when a CPU gets busy, or when the USB bus is under high utilization. Firewire also allows daisy chaining, eliminating the need for hubs in many situations.

Give me back my PS2 mouse ports and cordless mice, and fix the few Firewire issues that remain in XP/Vista and I'd be a happy man.

[I LOVE THE CAPS LOCK KEY @ Sep 19th 2007 11:31PM]

If you aren't using PS/2 then why does it bother you? Just ignore them and the BIOS should automatically disable them on POST every time. I have measured USB frame rate drops using a P4 2.4GHz and 7600GT. My second generation Logitech G5 set to 1000 reports per second can easily use 70% or more of my system resources with no games or applications running.

Firewire is better because it is a hardware solution. It is hardware based and does not involve the CPU for data transactions, It's also lower latency, and can operate in an isochronous peer to peer data exchange with no host computer required. In other words, I can connect a camcorder to an external storage device and instantly record to it. Very similar to bluetooth connecting to any other bluetooth device. Firewire rocks and is technically superior to USB. With that said, I hope USB 3.0 calls or allows for optional hardware accelaration like 1394 has.


Oh, and I forgot to mention; you can connect two computers together with a firewire cable and begin instant data exchanges with no need for a hub or networking gear. Firewire rocks!

[websyndicate @ Sep 19th 2007 8:41AM]

I use 6 Pin Firewire and love it. I would choose Firewire over USB anyday. Atleast with Firewire it doesnt make the processor work as much so there is less of a load when swaping data vs USB. I wish I had a Firewire 6pin Flash Drive

[I LOVE THE CAPS LOCK KEY @ Sep 20th 2007 11:41PM]

I saw it on engadget a while ago. OCz is going to offer a 1394 flash pen drive. I think newegg.com sold them for a while, but I may be wrong.

http://www.ocztechnology.com/aboutocz/press/2007/223
http://www.engadget.com/2007/03/16/ocz-shows-off-rally-2-firewire-thumbdrives-at-cebit/

[Andy @ Sep 19th 2007 8:44AM]

I want to know what idiot designed the USB plug shape ? Every single time I plug a USB device I seem to get it the wrong way round. OK - realistically, it's probably 50% of the time. But that's the point - it's almost impossible to tell which way to plug it in until you try it.

Firewire is much better in this respect

[Erwos @ Sep 19th 2007 8:55AM]

That's only true for 6-pin plugs. 4-pin plugs and 9-pin plugs are a little harder to distinguish in terms of "up and down".

I'd like to see Intel put some more development time into getting 1394c working, as someone else mentioned. A combined ethernet/1394c interface would kick serious ass.

[Scooter @ Sep 19th 2007 8:57AM]

Bang on observation.
Forget backwards compatibility, here is a chance to get the plug right. I once forced a USB plug in the wrong way and blew my motherboard. Try downsizing it too and ideally make it a universal shape (e.g. optical audio plug).
More power capacity would also be very welcome. The need for speed is a given.

[nextGen @ Sep 19th 2007 9:54AM]

"I want to know what idiot designed the USB plug shape ?"

Probably the same guy who designed the flat screwdriver around 1744.

[NewJohnny @ Sep 19th 2007 2:18PM]

I think Apple had it with the magnetic AC plug.

[PeterF @ Sep 19th 2007 4:00PM]

If you look closely, you can see that most USB plugs have the USB logo on the connector. For my monitor, the logo has to always face away from me, so I always get it right on the first try.

[Andy @ Sep 20th 2007 5:44PM]

What's really strange is that they got it right with the downstream end of the cable (a nice D shaped connector there), and the mini-USB cable (an asymmetric connector there). But the one you actually use most often is the hardest. A failure of HI design.

Yeah - flat screwdrivers frustrate me too, but at least there is no 'wrong way round' to insert the head into the screw !

[The Schizo @ Sep 19th 2007 8:47AM]

ok what now? 99% of USB2.0 pipes can't even fill themselves. If Intel is desperate to reduce bottlenecks, they need to work on HDD R/W. Even 2 RAID 0 SSD drives would (theoretically) only be able to sustain 120MB/s read and 80MB/s write.

http://www.samsung.com/eu/Products/Semiconductor/products/ssd.asp

[pinchies @ Sep 19th 2007 9:22AM]

Just want to note that USB 2.0 is theoretically 480Mb/s (which is 80MB/sec) not 480MB/sec.
Slight difference.

I also second what everyone is saying about processor load, and would like to see a "dynamic" voltage/current flow so all devices can be powered as efficiently as possible (particularly for mobile devices).

[The Schizo @ Sep 19th 2007 1:47PM]

*60MB/s, but the fact remains, it's still not the bottle neck problem here, it's the dang HDDs.

[Doug @ Sep 19th 2007 3:30PM]

60MB/sec is the bottleneck when transferring from a disk drive's cache memory, which happens a fair amount.

[timatl @ Sep 19th 2007 8:58AM]

4Gbps (300MBps) now that's what talking about. I am still very happy with USB 2.0

[mike @ Sep 19th 2007 8:58AM]

Has the author ever used firewire800? Data transfer via USB is painful compared to it. "Is firewire dead?"- are you kidding me?

Some of engadget's writers are reporting like cable TV news personalitis: all sensationalism, no substance.

[Thomas Ricker @ Sep 19th 2007 10:35AM]

@mike,

It's not sensationalism, Firewire ports are disappearing from CE class products. Sony and Apple started out as Firewire's most visible supporters and are slowly dropping it (iPods, handycams) or scaling back its use in their product lineups. This article is about a future spec and is thus forward looking. Of course FW800 is faster and less CPU hungry than USB 2.0 is *today*. But anyone who follows CE knows that the best technology often doesn't win. The industry is gravitating around Intel and USB and that is indeed a bad sign for Firewire.

Thomas

[mikeE @ Sep 19th 2007 11:02AM]

Yeah, you're right-the best technologies do too often fall by the wayside. Apple's discontinuation of FW support for the iPod was a tough blow.
I'm just a little bitter about engadget's "slaughter" of the Foleo- I was one of the 14 people in the world who was thrilled about it.

[Maff @ Sep 19th 2007 9:10AM]

when the ipod had a firewire cable included that was mint, it had a nice connector and was always faster than usb 2.0 on my compy

[g-vance @ Sep 19th 2007 9:22AM]

It's sad that it's gotten to the point where I have to read the comments on EVERY article to find out what's really the case. This place is getting sloppier and sloppier.

[Chris @ Sep 19th 2007 9:27AM]

Like many have pointed out, USB's problem is that their speed is *theoretical* and not even close to actual.

When I hook a 500GB external hard drive up to my computer with USB 2.0 and transfer files, I notice an extreme decrease in transfer speed than if I hook the same external hard drive using Firewire 400.

*Theoretically* the USB 2.0 should be faster but the Firewire 400 outperforms on a consistent basis.

I have no reason to believe USB 3.0 will be any different. So I'm sure Firewire is just fine.

[Dan @ Sep 19th 2007 9:31AM]

So does this possibly mean we might see webcams finally with better resolution and frame rates? If this technology doesn't help improve others, then it seems pointless, because Firewire does just as well for high speed transfer.

[DrNeuralgia @ Sep 19th 2007 9:43AM]

isn't that (webcams) supposed to depend on internet transfer speeds?

[adam @ Sep 19th 2007 9:33AM]

At 8 bits/byte, how is it that 4Gbps = 300MBps? Shouldn't that be 500MBps?

[trendless @ Sep 20th 2007 9:59PM]

Hear hear! Surely Intel's engineers wouldn't have messed up such an easy computation... :|

[Rik @ Sep 19th 2007 9:44AM]

Extra pin? I don't think that's a good idea, you have special propriatary interconnections for high speeds (I.E. eSata). Keep USB 3.0 pin compatible, make it faster and try to increase the power it can transfer. (Maybe thicker, certified cables).

[Eric @ Sep 19th 2007 10:12AM]

Umm... until USB is a any-any connectivity ( rather than host arbitrated like it is now ) it will still SUCK!

They call it a USB "hub" for a reason, only one device can talk at a time and it must be talking with the CPU. Devices can not talk with other devices without going through the CPU.