back to article IBM goes elephant with Nehalem-EX iron

With so much of its money and profits coming from big Power and mainframe servers, you can bet that IBM is not exactly enthusiastic about the advent of the eight-core "Nehalem-EX" Xeon 7500 processors from Intel and their ability to link up to eight sockets together in a single system image. But IBM can't let other server makers …

COMMENTS

This topic is closed for new posts.
  1. Tom Fleming

    Spinning..

    Wow, lots of spinning in the first few sentences. It looks like IBM is trying to capture the larger-server business. Perhaps they expect it to be high-margin business.

    What I don't get is how-well Windows (or Linux) will leverage all the additional CPUs and threads. Hopefully, there will be benchmarks soon.

    1. Anonymous Coward
      Happy

      re: Spinning..

      The benchmarks are amazing. IBM understands the trick is to keep the CPU's busy so they have built a balanced systems which allows for amazing memory throughput as well as a well thought out Solid State Drive design.

      ftp://ftp.software.ibm.com/eserver/benchmarks/news/newsblurb_performance_x_overview_033010.pdf

  2. Anonymous Coward
    FAIL

    "Plenty of RAS advantages"

    Enough of the "plenty of RAS advantages" press-release-fodder. Kindly put up or shut up, preferably with justified examples, and with particular reference to AMD64/Opteron rather than Intel's 2nd-rate clones. I'm sure Matt will be delighted to help you on the Itanic front.

    As far as I can tell, any substantial differences in RAS between an AMD64 enterprise-class Proliant, a mainframe, an Itanium, or whatever come largely from the OS software and associated management tools these days, and possibly from the kind of people working with them, and that isn't necessarily anything to do with The Chip Inside. E.g. a decent AMD64 Proliant with the NonStop OS on it (if such were to happen) would have pretty much the same RAS features and benefits as any other NonStop box, they'd hardly ever FAIL.

    I'd be delighted to be corrected if someone out there has *evidence* (not just unsubstantiated quotes from press releases and snippets from off-the-record vendor briefings) to show otherwise, but I've been waiting for "evidence" for years and not seen anything worthwhile.

    Have a nice long weekend. Thank you.

  3. Brian Miller

    2TB memory?

    Yoiks! Windows Enterprise and Data Center editions support that much memory, but I wonder if it is supported efficiently as Linux or Solaris. I've been told that the Windows kernel isn't optimized for multi-CPU systems. In one of the past Top500 lists an institution moved from Windows to Linux and nearly doubled the performance of their hardware.

  4. Anonymous Coward
    Anonymous Coward

    Minor correction

    The two way system is not called the 3960, it's called the 3690

    http://public.dhe.ibm.com/partnerworld/pub/misc/eX5_data_sheet_xsb03013usen.pdf

  5. Anonymous Coward
    Anonymous Coward

    OT: "an institution moved from Windows to Linux"

    You mean like the London Stock Exchange did recently, for example? No great surprise really, the bigger surprise is that the person in charge at the time the Windows decision was made is no longer in charge (you don't often see senior management in the City being held accountable for their foulups). What kind of idiot could ever imagine that Windows would be capable of handling the data rates and latencies that the madness called "high frequency trading" needs? HF trading is now responsible for maybe half the volume of trades around the world, and WIndows inability to cope is part of why the LSE has been significantly losing ground to its competitors, and the LSE ended up having to buy an outfit whose business is built around Linux-based trading systems.

    However, the relevant RAS in this picture is the RAS of the OS (Windows = rubbish wrt RAS, whatever Billco and his paid friends may say), whereas the RAS in the original article is *allegedly* hardware RAS.

    http://blogs.computerworld.com/london_stock_exchange_to_abandon_failed_windows_platform

    http://www.computerworlduk.com/technology/networking/messaging/news/index.cfm?newsId=18429

  6. Kebabbert

    A.C

    I know that NASDAQ runs it big stock exchange system on x86 and Linux. But I also heard that during the 9/11 attacks, the financial systems in the Twin Towers which ran on OpenVMS Itanium never dropped a single transaction due to OpenVMS' clustering technique. I do know that Itanium OpenVMS is held in very high regard in finance. Maybe if HP could port OpenVMS to x86, then we could see if there would be any differences in RAS? Recently Microsoft Windows abandoned Itanium. If this continues, then HP must port OpenVMS to x86, or it will die. And OpenVMS is too good to die.

  7. Anonymous Coward
    Thumb Up

    "OpenVMS is too good to die."

    "OpenVMS is too good to die."

    Anybody that's ever specified, designed, built, run, worked with, or (as yourself) read about and understood the value of a VMS system knows the truth in that (and it's *VMS* that brings that availability, Itanium as such brings nothing to the general availability picture when compared with a half decent "enterprise class" x86-64 box, or (I assume) half decent Power or SPARC box).

    HP management, and Compaq before them, do not belong to the categories of people that are interested in (let alone experienced in) VMS, and they do not listen to the people in those categories.

    Hence VMS in recent years has been distinctly unhip and largely invisible outside the world that already knows (and generally loves) VMS.

This topic is closed for new posts.

Other stories you might like