I stopped by the USB Implementers Forum booth, where they had both an interesting non-final demonstration of the new USB Power Delivery specification and some news about the new 10 Gbps USB 3.0 development specification. First, the 10 Gbps USB 3.0 development announcement stipulates continued use of the same cables and connectors for higher throughput, and although there will be a 10 Gbps certification program for cables the USB-IF expects almost all but the lowest end cables to pass without any reworking. I'm told that the newer revision enables 10 Gbps by using more efficient coding and also enables improved power efficiency. 

The standard is in development presently and should be completed by the middle of the year. It will then take the normal development cycle for silicon and controllers to make it onto the market. The USB 3.0 specification stipulates a meter long cable that's entirely passive, as opposed to other active solutions on the market presently. Longer lengths will require active componentry. 

I also wasn't aware but earlier this year the USB-IF finished the SSIC (SuperSpeed Inter Chip) specification which is essentially the USB 3.0 analog of HSIC (HighSpeed Inter Chip - USB 2.0). This is hugely important for smartphones and tablets where the limits of HSIC are already being encroached upon or actively hit with certain 2x2 802.11a/b/g/n combos, and will be a problem with 802.11ac combos. In addition I fully expect to see future SoCs implement SSIC for use with both these 802.11ac combos and faster 3GPP Release 10 or Category 4 basebands where there is a lot of attention as throughput continues to increase. 

Next up was a demonstration of the USB Power Delivery (PD1.0) specification working in conjunction with display driving. The USB-IF modified a Lenovo X300 notebook to accept power through USB 3.0 and the PD specification, and used a power delivery mechanism on the middle display to both power it, and the notebook with 65W. At the end of the daisy chain was a thumb drive plugged into the last monitor's hub which was available to read on the notebook. 

The X300 was running Vista oddly enough, which took a while to enumerate and connect to the other two displays, but did indeed simultaneously drive both a 1080p and 2048x1156 display through a daisy chain with good fluidity for a non-final demo. This combination of specifications is clearly USB-IF's response to Thunderbolt, and looks like a possibly even more impressive solution thanks to the ability to simultaneously drive the displays and charge the notebook over one cable.

Source: USB-IF (Power Delivery) (USB 3.0 10 Gbps)

Comments Locked

26 Comments

View All Comments

  • Aikouka - Sunday, January 13, 2013 - link

    As I sat there reworking the cables behind my TV stand last night, I kept wondering why we don't attempt to consolidate cables more often. The HDMI spec supports an Ethernet channel, yet I'm still running CAT5e to all of my devices! Even if it's just removing a single type of cable, it would bring me one step closer to cleaning up the clutter! Imagine if I could run a single cable from my AVR to my XBOX that provided power and Ethernet while also serving the video/audio from the XBOX to the AVR! =)
  • praeses - Sunday, January 13, 2013 - link

    I think you have it a bit backwards, you would be better off running everything over cat 5e/6 instead (as the connectors, cables, etc are much cheaper/ubiquitous). There isn't much that can't be run over "ethernet" cable including: usb, vga, hdmi, dvi, power, sata, video cameras, alarms, phones, heck there's even pci-e over cat 6.
  • toyotabedzrock - Sunday, January 13, 2013 - link

    I think Displayport and HDMI are capable of much higher bandwidth. Remember they handle raw decoded data not an mp4 file.
  • defter - Sunday, January 13, 2013 - link

    The latest version of HDMI supports 8.16Gbps of real bandwidth. Therefore 10Gbps Ethernet that works over cat 5e/6 is more than enough.

    The only reason for HDMI is ripping off consumers and limiting competition (HDMI support requires license payments that small companies may not afford).
  • Guspaz - Monday, January 14, 2013 - link

    While Cat5e and Cat6 cables sort of support 10GigE, even the official standard limits their range to well below the normal levels, even in perfect conditions. Interference is a big enough problem that Cat5e cables often have trouble with regular gigabit ethernet, let alone 10GigE. It's unlikely that, shoving ethernet cables in the rats nest behind a TV, you're going to have success with anything less than Cat6a, which isn't cheap. Of course, there's also the issue of 10GigE NICs being way too expensive for consumer use...

    You're also a bit off on the specs. 8.16 Gbps is the throughput after removing overhead. With overhead, they're at 10.2 Gbps, and HDMI 2.0 is expected to bump that up to 18 Gbps.

    HDMI's ethernet support is also limited to 100 Mbps, which limits the usefulness somewhat.
  • GTRagnarok - Sunday, January 13, 2013 - link

    USB powered laptops will be awesome.
  • toyotabedzrock - Sunday, January 13, 2013 - link

    Yes I'm looking forward to all the broken connectors because it will be much easier to solder a USB port.

    </sarcasm>

    Why hasn't anyone created a magnetic or otherwise unbreakable solution for PCs yet?

    I have several ideas on how to not violate their patents.
  • Zink - Sunday, January 13, 2013 - link

    I have my laptop down to three cables when I plug it in: USB hub, display, and power. This almost perfectly brings everything over one cable. The big downside is USB display lacking any graphics acceleration but most users wouldn't mind that. Thunderbolt is the technically superior solution to get high bandwidth and graphics acceleration but USB will hopefully be the cheap alternative.
  • amb9800 - Sunday, January 13, 2013 - link

    The big advantage of Thunderbolt is that it actually passes video straight over the cable, whereas with USB you need a separate display adapter (here they're using DisplayLink). In the video, it seems the left monitor might have a DisplayLink controller built in, and the right monitor has a controller taped to the back.

    DisplayLink has advanced quite a bit, but it's still not a native display connection. I'm actually driving a 27" 1440p LCD via DisplayPort from an HP USB 3.0 port replicator that has a DisplayLink chip inside (since HDMI 1.3 and MicroHDMI can't do more than 1080p). Performance is not bad with a high-end quad-core i7 laptop, but power users can tell that something's up-- and when connected to my i7 ultrabook (not exactly a $200 netbook), things start getting messy (feels like using an RDP connection to some PC).

    It's a workable hack today (given so many PCs still lack DisplayPorts and are thus unable to drive >1080p displays), but not really comparable to a real video connection.
  • NeBlackCat - Sunday, January 13, 2013 - link

    It blows my mind that USB hasn't been standardised as a native display interface yet.. Would be great for running headless low end servers, rescuing laptops with broken screens, etc. Just have a small USB connected monitor on hand. Better yet, imaging if you could boot into a laptop's BIOSm enable an "Act as Monitor mode" and use it as a monitor via a USB device port on the back. Lovely, and mainly a software effort.

Log in

Don't have an account? Sign up now