avoton

Latest

  • Intel introduces next-gen Xeon E7, E5 and E3 families for enterprise space

    by 
    Richard Lai
    Richard Lai
    04.10.2013

    Truth be told, not much has been announced on the consumer end at IDF Beijing earlier today, but Intel did tease us with upcoming refreshes of its Xeon E7, E5 and E3 families for the enterprise space. In chronological order we have the Haswell-based E3 with TDP as low as 13W, and it's coming in mid-2013. This is followed by the "Ivy Bridge-EP" E5 in Q3 and then the "Ivy Bridge-EX" E7 series in Q4, the latter of which boasting three times the memory capacity of its predecessor, along with Intel's Run Sure reliability feature. More details in the press release after the break if you're into these flavors of chips. Also mentioned at the keynote were the now-available Atom S12x9 family for storage systems, as well as a couple of upcoming 22nm 64-bit Atom SoCs codenamed "Avoton" and "Rangeley," both of which are sampling now and are expected to launch in the second half of this year. We've actually already heard of the microserver-friendly Avoton from Facebook's Open Compute Project, whereas Rangeley for network infrastructures was also detailed around the same time; so again, hit up the press release for more details. %Gallery-185168%

  • Facebook's Open Compute Project splits up monolithic servers with help from Intel, more

    by 
    Jon Fingas
    Jon Fingas
    01.16.2013

    As much as it's important to have every component of a PC stuck together in a laptop, that same monolithic strategy is a major liability for server clusters: if one part breaks or grows obsolete, it can drag down everything else. Facebook and its Open Compute Project partners have just unveiled plans to loosen things up at the datacenter. A prototype, Atom-based rackmount server from Quanta Computer uses 100Gbps silicon photonics from Intel to connect parts at full speed, anywhere on the rack. Facebook has also garnered support for a new system-on-chip connection standard, rather affectionately named Group Hug, that would let owners swap in new mini systems from any vendor through PCI Express cards. The combined effect doesn't just simplify repairs and upgrades -- it lets companies build the exact servers they need without having to scrap other crucial elements in the process. There's no definite timeframe for when we'll see modular servers put to work, but the hope is that a cluster's foundations will stay relevant for years instead of months.