Here are two articles written by other bloggers then edited by me. If you are one of the other bloggers and do not want this content on this site you may email me at firstname.lastname@example.org to tell me. I will take your content off imeadiatly.
Intel announces, demonstrates USB 3.0
By Joel Hruska | Published: September 18, 2007 – 10:12PM CT
One of the announcements to come out of the IDF keynote this afternoon was Pat Gelsinger’s discussion and demonstration of USB 3.0 technology. Although still in the prototype stage, USB 3.0 is aiming for 10 times the bandwidth of current USB2.0 solutions, or approximately 5Gbps. Since this requires fiber optic cabling, USB 3.0 will add a length of optical data cable to the mix, though USB 3.0 will retain full compatibility with USB 2.0 (and, one assumes, USB 1.0 as well).
Gelsinger expects the final version specifications to be finished by the first half of 2008, with USB 3.0 peripherals realistically appearing in 2009 or 2010. There are an increasing number of devices that could potentially take advantage of the additional bandwidth, including external hard drives, flash readers, video cameras, and the all-new USB-compliant llama expected to be genetically delivered from the Andes sometime in late 2011. Obviously a USB llama wouldn’t be much use without a USB 3.0 device—ever tried downloading from a llama over an old 10BaseT network? Yeah. It’d be worse.
As for the other, official features of USB 3.0, there remains quite a bit of information we don’t know, and it would have been nice for Intel to have included additional information. USB has long been criticized for relatively high CPU usage. This has inevitably become less of an issue as CPU performance has improved, but devices capable of using USB 3.0’s higher bandwidth capabilities could make CPU usage a problem again unless the issue is addressed during spec development. Issues like cable length, available power provided, and the number of devices per channel are all unrevealed as yet, and possibly unresolved. The Inquirer has a few more details on the spec (and the rest of the keynote) if you’re interested.
As far as future market competition, its target of 5Gbps puts USB 3.0 ahead of current eSATA (3Gbps), which is really the only other device protocol under active development that might challenge it as a peripheral interconnect. Although an IEEE 1394c protocol has been developed and published as of June 8 2007, no company has announced an intent to produce a product or chipset that utilizes the standard. FireWire remains supported in certain sectors, but I’d personally be surprised if the combination of USB 3.0 and eSATA doesn’t push FireWire out of the market completely. As for the nascent specification, the proposed 5Gbps speed is great and all, but hopefully the development committees will acknowledge some of the other concerns regarding the USB 2.0 protocol and incorporate solutions for them, rather than carrying them over into another product generation.
Gelsinger demos USB 3.0, PICe 3.0 and other new toys
Intel Fall 007 Kicking ass
By Charlie Demerjian in San Francisco: Tuesday, 18 September 2007, 10:46 PM
PAT GELSINGER GAVE the afternoon keynote at IDF today, and it started off with some rather tame enterprise parts, then moved on to the fun stuff. First up was virtualisation, and a show and tell of the virtualisation vendors.
The one interesting one was presented by John Fowler of Sun with its upcoming, unannounced VM on its unannounced Tigerton/Caneland box. The most interesting bit is that, because it is running on Solaris,and you run Windows under that, the IO and storage are Solaris based and simply exposed. If you run Windows under Solaris, you get a real fault-tolerant file system on a toy OS.
The one time Itanium was mentioned was in a new Hitachi Virtage blade. This blade is running Montvale, aka Montecito v.07. Other than that, the red-headed step-chip was absent from just about everything.
Next was security, and they gave the quick rundown of VPro 2007 aka Weybridge. We told you all about it here, check it out if you are into virtualisation security. The next-gen VPro is called McCreary, coming in 2008. It will have a TPM v1.2 on the chipset, AMT 5.0 and a technology called Danbury. We will have a full write up on Danbury later today.
Going to I/O, we have the announcement of PCIe 3.0, Quickassist and USB3.0. PCIe 3.0 has twice the bandwidth – that would be 10Gbps – dynamic power management and supports accelerators. This used to be called Genesseo, but now has some marketing name related to Quickassist.
USB3.0 is probably going to be the biggest one of the bunch. They are aiming for 10-times the bandwidth, which would put it at about 5Gbps. For this you need optical, and USB 3.0 cables have an optical link in the current form. Backwards compatible, loads of bandwidth, optical and hopefully available in 1H/08.
Another optical link was the FCoE announcement, aka T11. It does just what it sounds like; puts fiber channel over vanilla ethernet. Intel has 10GigE adapters in copper and fiber for that, buy 12.
Then comes storage. Intel is getting into SSDs and they were showing of prototypes and vague specs. The specs were SATA 3.0, 10-50x the IOPs and a 4.5x power savings, all with twice the write speed. What this is over was not stated, but I don’t think it will be all that slow in any case.
Gelsinger then mentioned Skulltrail and pointed to a box without any more comment. There are several of these boxes at the show, so more info will be forthcoming.
The system diagrams for Nehalem were also shown off, but that is old news, see here and here. They then demo-ed probably the most important bit of the show, a 2S Nehalem system running. Getting one CPU up is easy, the second is much much harder with a new interconnect. Intel did it.
In general, there was a lot new here, more than can be covered in a single story. Much of it will be on your desktop in a year or two, most of it not esoteric server side technology. All I have to say looking back to the spring IDF is …. Yo! µ
My initial reaction to the new USB is wow 5 gb!!!!! Holy shit that’s a lot of data to go in one second. Then I thought about it and realized: Ya that’s fast but that number is not the actual speed you’ll get it is the original data throughput. Also I don’t think they will be able to deliver something quite that fast right away. From just over 400 megs to 5 gigs is a long way to jump overnight. I must admit that no matter what it’ll be fast and well worth installing in a machine. It’s just questionable weather they will actually deliver something as good as they say it is.