Unveiled on Monday by the USB Implementers Forum, the USB 3.0 spec can theoretically support data-transfer speeds of up to 4.8Gbps – 10 times the speed provided by USB 2.0. The new standard, also known as SuperSpeed USB, is also expected to be more power-efficient than its predecessor. “SuperSpeed USB is the next advancement in ubiquitous technology,” Jeff Ravencraft, the president of the USB Implementers Forum (USB-IF), the industry group that promotes USB technology, said in a statement on Monday. “Today’s consumers are using rich media and large digital files that need to be easily and quickly transferred from PCs to devices and vice versa. SuperSpeed USB meets the needs of everyone, from the tech-savvy executive to the average home user.”
“USB 3.0 spec can theoretically support data-transfer speeds of up to 4.8Gbps”
And the speed in reality?
USB 2.0 can support 400MBps, but most hardware manufacturers are incredibly cheap in product design. I see 12MBps out of most modern usb devices, save for some relatively expensive devices.
the speed in reality will also depends on how fast the data can be read and written to ..hard drives rps will matter, under what stress the os is under will matter, how much fragmented the hard drive will matter ..
as they say, “you are as fast as your slowest component” ..to total speed will depends on a lot than just what the spec says ..
Hold the train for a second. I think we’re getting seriously confused on terminology. Whether or not the B is capitalized makes a factor of 8 difference! bps is bits per second. Bps is Bytes per second. There are 8 bits to a Byte. USB 2.0 supports a theoretical 400 Mbps, therefore a theoretical 50 MBps. So, if you’re seeing 12 MBps this is on the order of 25% of the theoretical maximum, and makes good sense.
USB 3.0, if we are to believe the titlepage is correct, will support 4.8Gbps, so in other words, somewhere around 600MBps range.
Please note, this is further confused by the fact that 1 GB != 1000 MB, but rather 1024 MB, so please realize all these numbers are approximate. To be honest, I’m not sure if the manufacturers report using 1000 bits/Kbit to make things look faster, the same way hardrive manufacturers do to make drives look bigger.
edit: typos!
Edited 2008-11-17 21:17 UTC
No one is confusing Bits for Bytes. We’re pragmatic in pointing out the lie that is “Theoretical” vs. “Actual” throughput.
I’m looking forward to seeing the sustained [non-burstable] throughput of USB 3.0 versus Firewire 3200. I know FW’s peer-to-peer model is superior to the master-slave and I’m not expecting USB 3.0 to surpass FW3200.
That however doesn’t change the fact that USB is the standard due to INTEL owning and controlling this particular market with their USB standard.
Realistically though, with overhead of the protocol (control bits, etc.) a serial bus throughput usually boils down to a fraction of 10 anyhow.
Thus, 480Mbps (the actual alleged throughput of USB2, not 400 as was indicated) usually ends up with ~48MB/sec maximum… I’m not sure I’ve ever seen it hit this amount anyway
I was thinking that too, it isn’t 400Mbps, it’s 480Mbps.
Back when USB 2.0 was coming out, I had a discussion (really more like an argument telling a person how stupid he is) about USB 2.0 and Ultra SCSI 160.
I believe his exact words when I told him I was building a system with Ultra SCSI 160 drives, “you’re jumping on the bandwagon a little late, USB 2.0 is going to crush SCSI.” Yeah, and when did that happen?
He failed to understand that 480Mbps is not the same as 480MBps, and that 160MBps throughput of the Ultra160s would crush USB2.0
>And the speed in reality?
Fast enough to torpedo FireWire ?
Edited 2008-11-17 20:31 UTC
Not when USB isn’t a real bus and wastes CPU, is horrendously inefficient and can’t do half of what FireWire can. USB will always be second fiddle, quality wise, when we have FireWire and eSATA.
the new USB spec is actualy rather good (though i preffer the new FireWire spec hands down). as far as eSATA, it will only become awsome when the revision is finalized that will power the drives over the eSATA cable. but thats still a little ways off last i checked.
And yet there are many, many, more storage devices using USB than Firewire and eSATA combined. It might not be ideal (I use an eSATA external disk myself), but it’s cheap, and it works well enough for most people.
True. USB consumes noticeably more CPU and requires a host computer (FireWire can communicate without a host). FireWire also has modes that USB does not. This makes FireWire more suited for real-time applications.
Note also that FireWire is an IEEE standard (IEEE1394, with various sub-versions). It is also a US military standard which has led FireWire to be used on the F-22 and F-35 fighters. IIRC, NASA also uses FireWire in equipment on the Space Shuttle. USB is an ‘industry standard’ maintained by a consortium of interests, which is why there are some dodgy implementations out there (and why it was possible for some USB 1.1 devices to be labelled “USB2” when they were not).
USB is far better than the mess of connectors that came before but technically it is not as good as FireWire, so don’t believe the marketing bullsh!t. It was principally designed to be a “poor man’s FireWire” mostly so that Intel and Microsoft wouldn’t have to pay royalties to Apple for all the devices they envisaged.
Well think of all the cool new things like better gaming mouses?
Well why would AMD fork USB im actually surprised that AMD have not released a external device on HT yet. I think its good that we are going to see USB 3 shortly because those external hard drives on USB 2 still suck.
Firewire flash never took off (is there any made?)
Still its nice to see that such important device specifications arnt being forked. DVD +-R DVD RAM… blah blah omg why so many standards isnt two enough?
It is also important that there is another standard to be competitive with the other and both have different kinds of market share. Sadly USB has won once the MAC removed firewire as its default.
Exactly what I was thinking. Apple, the inventor of FireWire, has been slowly de-emphasizing its use. The new MacBook model does not include one and I suspect that slowly the iMac and other models will soon be popping up without them too.
As to the never-ending debate as to which is superior USB or Firewire, that is going to wage on long after FW is gone. Firewire is technically more capable, but is also more expensive to make and requires more logic then USB. FireWire devices have all but shriveled up to nothing and it is dying off fast. Also USB bridgeboards for HDs have improved drastically over the years too, to the point that they are often as fast or faster then their firewire counterpart. I did a series of benchmarks a year or so ago to see which one was faster transferring the same files from the same drive and same computer. The USB interface won out over the two firewire ones I used, one an oxford chipset and one a generic. So for most people they are not going to care about FireWire as USB will continue to dominate and take over.
Would this also be an option to connect a monitor to the PC?
You can do this already:
http://www.cooldrives.com/usbvicausbdi.html
for example
1. As others pointer out, we are talking about Gbps (Giga-bit-per-second) which translates to ~600MBbps. (MB/s).
2. As far as I remember (am I?), USB uses the same 8B/10B encoding used by PCI. Drop the 600MBps by another ~12%. dropping the theoretical maximum bandwidth- to ~490MBps.
3. At least in my experience, USB<->host interface is -very- CPU intensive (Compared to PCI-X/E). Unless something fundamental changed in USB3, I doubt that someone will max out the theoretical bandwidth offered by USB3.
Never the less, having said all that, USB3 is very promising indeed.
… Now if only they offered a better solution to the power-over-USB problem (read: USB-power can only power small devices due to design limitations), I’d be a very happy man.
– Gilboa
Edited 2008-11-18 10:58 UTC
So now we have:
(Low Speed)
Full Speed
Hi-Speed
SuperSpeed
I wonder what’s next… UltraSpeed? HyperSpeed? Why those silly names… it always seems like there’s nothing more to improve, so they’ll always have to pick a new name that is even more ultra than the old one.
Well, isn’t it obvious? LUDICROUS SPEED!
“They’ve gone to plaid!”
Too much Space Ball here! LOL
Why the hell do we insist on running power over data lines? I work for a large IT university department and the number one cause of all external drive failures is that the power shorted out the chipset. Its just plain stupid. I would much rather have an external power standard than try to run it over the data lines
EDIT: Thats also assuming that it didnt take the whole motherboard with it.
Also, esata stomps USB/firewire. I get native speed from my external drive, same as if it was internal, 110MBps. All that with 0 cpu hit. Plus eSata allows for port multipliers allowing 5 devices per port. That is a bottleneck now, but once devices have finished switching over to PCIe 2.0, a port will handle 500MBps throughput.
Edited 2008-11-18 18:18 UTC
The USB-IF hopes USB 3.0 will be built into computers from late 2009, with consumer products using the specification starting to appear the following year
I would think computer manufacturers would start to put them in within 6 months of the specifications being certified. Is there something about its implementation at the hardware level that is particularly difficult or is the OS the “bottleneck”?
If version 3.0 is SuperSpeed USB
Then version 4.0 would be HyperTurboSpeedRicer USB
or something like that…