On Sat, Jan 16, 2010 at 03:59:12PM -0500, Alan Stern wrote: > On Sun, 17 Jan 2010, Sarah Sharp wrote: > > > The USB 3.0 specification does list the speeds in an interesting way > > that makes me think we could see USB 3.0 devices be faster than 5Gb/s. > > Checkout Table 9-13. > > What's special about this table? And how does it indicate you could > get speeds faster than 5 Gb/s? Bits 0 through 3 all mention the name of the speed, i.e. low speed, full speed, high speed. Logically, bit 4 should say Super Speed. Instead it mentions 5Gb/s. And they have plenty of reserved spaces after that to add speeds that USB 3.0 devices can operate at. I've heard talk that some people want to rev the speed without reving the USB 3.0 bus specification, but that's the only physical proof I have. > (Actually, 5 Gb/s is a physical speed, not a logical speed. Because of > the 10/8 encoding, each byte occupies 10 bits on the line. Hence the > maximum logical speed is 500 MB/s, rather than 625 MB/s.) > > I have to admit, though, the description for the wSpeedsSupported field > is a little strange. It says: > > Bitmap encoding of the speed supported by this device > when operating in SuperSpeed mode. > > Which is ridiculous, because when it's operating in SuperSpeed mode the > only speed it supports is SuperSpeed. I believe the SuperSpeed USB Device Capability descriptor shows up even when the device is connected under EHCI. So Windows (or Linux) could pop up with a message saying, "This device would work faster under xHCI." Alternatively, when bandwidth is limited, the USB core could force some USB 3.0 devices down to a slower speed. I forget exactly how you can do this with the xHCI driver, but I have vague memories of discussions about it. Sarah Sharp -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
