USB 3.0 gets seriously detailed
By Darren Murph 
posted August 18th 2008 3:36PM
posted August 18th 2008 3:36PM
posted August 18th 2008 3:36PM
A look back on popular stories from today in a specific year.
Now that we've thrown 'em off the trail, use the form below to get in touch with the people at Engadget. Please fill in all of the required fields because they're required.
Richard: NO. I'm not going to say more than that or I violate agreements.
Neat!
What about wireless USB?
How much fast does this compare to lets say firewire?
To add my two cents to issues being discussed here
1> Yes the raw datarate is 10x not 5x USB2.0's as someone mentioned
2> Higher power means you can still use it as a (better) kettle because the power consumption savings are for when you're NOT utilizing the device
3> @ethana > Try designing one of those protocols that autoscale like you mentioned. At these speeds, its hard enough to get fully synchronized with tightly controlling cables, hubs and controller/PHY specs. The autoscale IS implemented, you can connect at USB 2.0/1.1 speeds as well as 3.0 speeds.
4> for everyone who has trouble with a connector, please look at the connector holes like someone mentioned. It is about as easy as using an escalator (which means there is some room for people to screw it up, as I have seen).
Look at the holes and do what exactly? Its impossible to know which way to plug them in without trial and error. Not all manufacturers have their USB ports facing the same way, my Dell monitor takes them one way, the back of my motherboard takes them the other way, both are vertical. The front of my computer and my PS3 have horizontal ports but dont take USB things the same way. All the devices are different, the holes mean absolutely nothing. Its a horrible design, probably the most frustrating port of all time.
consider me schooled :)
@chefgon_ign :
Sorry to hear about your wife getting a new "Computer." What are you doing now that you are missing your little connector that floats out in the middle?
There's a couple of observations that I have with this announcement.
1) "90% complete" still isn't 100%.
As per the Pareto Principle, they've gotten the easy stuff done, because its the final 10% that is where the really hard work lies. As such, this is still has a strong scent of Vaporware.
2) The USB plug ... and "backwards compatible" bit ... are problemmatic.
As has been pointed out here already, the design of the current plug is a Human Factors design failure - - as per Murphy's Law, there's a 99% chance of landing 'Peanut Butter Side Down' when trying to make a simple connection. The simple reality is that the "holes" are a small band-aid on a bad interface design. This design flaw could have (and IMO should have) been corrected with the USB3 design, but in trying to compete with FW3200, they're trying to play up this "plug compatible" card, even though it is highly disingenuous: the facts of the matter are that if you hook up two USB3 devices with a USB2 cable, you'll only get USB2 bandwidth.
Having USB2 and USB3 cables have interchangable connectors - - but not actually be interchangable - - will inevitably cause consumer confusion and frustration. The lesson-NOT-learned here is that a "downward compatible" is not the same as "interchangability".
3) Theoretical I/O numbers don't mean Squat.
Please remember that despite its "higher" specifications, in the real world, USB2 has around 35% lower bandwidth than Firewire 400. The basic reason for this is due to how much of the available bandwidth is consumed by overhead, which is what ultimately determines the effective I/O efficiency of the protocol.
For example, for USB2's "480" to be ~1/3rd slower than FW400, this means that if FW400 were theoretically 100% efficient (it isn't), then the MOST efficient that USB2 can be is around 60%:
(400 * (1-.33))/480 = 55.6%
Even if one wants to argue that the bandwidth of USB2 and FW400 are "exactly the same" (as opposed to the reality of FW being better), then the maximum theoretical efficiency of USB2 is:
(400/480) = 83%.
What is constantly missing from these splashy articles is the real world perspective of "Good, but how fast does it REALLY go?", in the real world of benchmarks and technical analysis of interface protocols.
I've tried to find published citations on the bandwith overhead consumed by Firewire and USB, without that much success. The best that I've personally found is that FW's overhead is 3% and that USB's overhead is 40%.
Applying these (wobbly reliability) values to the inevitable FW3200 -vs- USB3 shootoff, we get:
FW3200:
3200 * (1-.03) = 3.104 Gbps
USB3:
4800 * (1-.40) = 2.880 Gbps
Gosh, USB3 is ~7% slower...?
Of course, this is just one prediction, based on bandwith overhead number claims of wobbly reliability. Nevertheless, we do know that no I/O protocol can be 100% efficient and we know that USB's overhead is not insignificant. One of the questions is going to be by how much can it be improved while retaining backwards compatibility? As I said, there's "HARD WORK" still remaining to be done in that last 10% of the USB3 Standard.
The proof will come about after both Standards are finalized and shipping, so that no-kidding real world performance benchmark tests ... on both ... can be conducted in an objective and professional manner.
-hh
i hope it can as fast as it said! so cool~
wow this is so Gooood
Admin Please Come to my Blog too :
http://www.WarezHouse.info