First USB3 Host Controller Now Available
Cue applause.
The host controller supports up to two USB 3.0 devices, and connects to the northbridge via a PCI-E Gen 2 link, as shown in the diagram below.
And hey—if diagrams are really your thing, we've got this:
What we've got here is a USB controller with two external USB 3.0 ports. If a USB device is plugged in that doesn't support USB 3, there's an entire USB 2.0 implementation already on-chip. On a happy note, the presence of a separate USB 3 controller means two different devices can hook into the same port without both defaulting down to USB 2. Both controllers hook into Intel's xHCI (Extensible Host Controller Interface), which in turn attaches to a PCIe Gen 2 x1 interface. At that point, you drop the card in an actual PCIe slot on your motherboard, and everything is good to go.
In theory, this is actually a less-than-ideal situation. A first-generation PCIe x1 slot has a maximum theoretical transfer rate of 200MB/s, Gen 2 improves this to 400MB/s. The single-lane connection should deliver more than enough bandwidth, regardless of whether or not its Gen 2, but the USB 3 bus's maximum speed of 4.8Gb/s means we'll have an external interface that's potentially faster than the internal port it attaches to. If you want to saturate USB 3, even on an x1 electrical connection, you're going to need some serious mojo.
Now all we need are devices that use it.
|
Via: Business Wire | News Archive
| Tags:
Intel,
NEC,
xHCI,
USB 3,
USB 2,
host controller,
PCIe x1,
PCIe Gen 2 x1
|
Comments
|
... which will take another year or two to appear. |
|
"In theory, this is actually a less-than-ideal situation. A first-generation PCIe x1 slot has a maximum theoretical transfer rate of 200MB/s, Gen 2 improves this to 400MB/s. The single-lane connection should deliver more than enough bandwidth, regardless of whether or not its Gen 2, but the USB 3 bus's maximum speed of 4.8Gb/s means we'll have an external interface that's potentially faster than the internal port it attaches to. If you want to saturate USB 3, even on an x1 electrical connection, you're going to need some serious mojo." Indeed, but at least its good news for the future (now if only we can replace the aging PCI e's). |
|
It's in the works, gibbersome. PCIe 3.0 is supposed to be finalized in 2010, with products likely in 2011. Too bad the world ends in 2012. |
|
Well, PCI-e 2.0 X1 is actually closer to 500MB/s (according to the Wikipedia gods). And USB 3.0 is 4.8Gb/s which is only 600MB/s... so it's not as bad as it seems. Besides, I think only SuperSpeed USB 3.0 products can theoretically even achieve that range and not all USB 3.0 will be that (I think). You'll be fine with PCI-e 2.0. |
|
I'm already fine with USB 2.0 speeds and my Firewire 800 devices. All this "new and faster" isn't a bad thing, but when is enough, enough? My finances cannot keep up now that I'm retired. |
|
Crisis Causer: PCI-Express uses an 8b/10b encoding scheme. This means that an eight-bit data payload is encoded for transmission as a 10-bit payload. Yes, the bus is technically 10 bits wide, but those two "extra" bits contain no data and are considered overhead. A PCI-E Gen 2 x1 slot is capable of transporting data at up to 400MB/s. The missing 20% is irreducible overhead. |
|
Realneil, The real advantage of these next-gen interfaces, at least in theory, is that they'll give us external interfaces to pair with SSDs. A standard hard drive may not be crimped much by USB 2, but an external SSD hooked to USB2 would be hamstringing itself. |
Post a Comment
