back to article ARM plans to win 20 per cent of the server market by the year 2020

ARM Holdings is telling investors it will take 20 per cent of the server silicon market by the year 2020. The company made the claim last week at an event in Taiwan, sending ripples through the Asian media. A little searching shows that presentations like this one (PDF) shows that the 20 per cent claim has been used in chats …

  1. Christian Berger

    The problem is still the lack of a decent common hardware plattform

    I mean for servers most customers will want to run their own operating system. Trying to sell a server which only runs Ubuntu 15.4 or something is very hard as people will want to install whatever they want.

    Virtualization might ease the problem a bit, but they pose new security risks. If you have a vendor provided hypervisor you cannot easily replace, who will guarantee that it's free of any deliberate or accidental security holes?

    If there just was a simple architecture. Essentially a way which specifies how the serial console is connected to your ARM core, where your RAM is and how to access your flash memory and your PCI(e) bus, we'd already be a great deal further. That way manufacturers could build a great variety of different systems, while maintaining compatibility with others.

    1. Anonymous Coward
      Anonymous Coward

      Re: The problem is still the lack of a decent common hardware plattform

      "Trying to sell a server which only runs Ubuntu 15.4 or something is very hard."

      Yep - they will probably need Hyper- V Server for Arm and Windows Server for Arm to make any real progress in the Datacentre. Windows already runs on Arm though, so just a case of Microsoft wanting to support it on the server side...

      1. Richard 33
        WTF?

        Re: The problem is still the lack of a decent common hardware plattform

        Windows is pretty much irrelevant in serious data centers. All these ARM servers will be running Linux.

        1. Anonymous Coward
          Anonymous Coward

          Re: The problem is still the lack of a decent common hardware plattform

          "Windows is pretty much irrelevant in serious data centers"

          Most 'serious' data centres were historically running ancient UNIX OSs on midrange systems, these days the vast majority of which are being migrated to Windows. Just look at the SQL server market share gains for instance: http://www.cmswire.com/cms/big-data/microsoft-sql-server-wins-in-a-big-data-world-024565.php Sure a few go to Linux where they don't fancy updating dated legacy skilled staff, but it's a space where Windows is overwhelmingly winning. The only spaces where Linux has majority use are web hosting and HPC.

          1. Jim 59

            Re: The problem is still the lack of a decent common hardware plattform

            ...Windows is overwhelmingly winning

            This is an obvious troll comment, but I'll bite. A datacentre contains a goodly amount of both Linux and Windows, various other Unixes, and the occasional mainframe. Windows did exactly as you describe in the late 90s - muscled in on midrange at the cost of propriety Unix systems. But from 2000 on, Linux did the same to Windows.

            Nowadays the heavy /specialist work is often still done by proprietary Unix on proprietary hardware. The medium work is done by Linux on commodity hardware, and the light work (eg running basic network services to support the desktop) is done by Windows. The natural order has remained fairly unchanged.

            Oh and Windows is still king of the desktop, the king of embedded is Linux.

      2. Anonymous Coward
        Anonymous Coward

        Re: The problem is still the lack of a decent common hardware plattform

        "Yep - they will probably need Hyper- V Server for Arm and Windows Server for Arm to make any real progress in the Datacentre. Windows already runs on Arm though, so just a case of Microsoft wanting to support it on the server side..."

        Right. So for those who have the Microsoft blinders on, the strategy with ARM is to:

        1. Wait for MS to release a build of Windows Server for ARM.

        2. Wait for MS to release required Windows Server features for ARM.

        3. Wait for application vendors to release their products on ARM (while some I've dealt with are still figuring out x86-64).

        While you twiddle your thumbs, the rest of us who have actually diversified our skillset simply load up a *nix on our ARM hardware and get on with it. As an example, OpenStack already works on ARM, so there's a pretty good start right there.

        1. Anonymous Coward
          Anonymous Coward

          Re: The problem is still the lack of a decent common hardware plattform

          ARM servers aren't targeted at Windows, they're targeted at hyperscale stuff where lots of MIPS are important, and Linux absolutely dominates there with Windows having presence only in 100% Microsoft shops.

      3. Richard Plinston

        Re: The problem is still the lack of a decent common hardware plattform

        > Windows already runs on Arm though,

        Windows CE, Windows Phone, Windows RT run on ARM. Not quite server grade are they. But that is not particularly important, what MS would need is MS-SQL Server, IIS, Active Directory and much else to run on ARM.

        With Linux it is already there.

    2. CGX

      Re: The problem is still the lack of a decent common hardware plattform

      Something like this?

      http://www.theregister.co.uk/2014/01/29/arm_standardization_sbsa/

      But no, I don't know any actual implementations of the spec.

      1. Christian Berger

        Re: The problem is still the lack of a decent common hardware plattform

        "Something like this? http://www.theregister.co.uk/2014/01/29/arm_standardization_sbsa/"

        Yes exactly. Though nobody knows if they will succeed.

    3. Richard 33
      Go

      Re: The problem is still the lack of a decent common hardware plattform

      You probably shouldn't judge ARM by the wild west of low-end non-server 32 bit hardware currently available. 64 bit hardware should all conform to the SBSA, which means it does have a serial port, in a known location, and UEFI to boot and ACPI to define the hardware to the OS. I do think you have a valid point though. If "server" manufacturers ignore SBSA, ACPI and similar then ARM will fail for the exact reasons you outlined.

  2. DainB Bronze badge

    Intel Broadwell-D

    Bye-bye ARM server market, we heard so much about you but never had chance to see.

    1. Salts

      Re: Intel Broadwell-D

      Hmmm, I don't agree with you(though did not down vote you) but you should have posted

      Goodbye ARM server, Though I never knew you at all

      You had the grace to hold yourself... :)

    2. P0l0nium

      Re: Intel Broadwell-D

      I surely hope these "Investors" aren't managing MY pension pot :-)

      http://www.anandtech.com/show/8357/exploring-the-low-end-and-micro-server-platforms/18

      Which concludes :

      "It is unfortunate that AppliedMicro's presentations have created inflated expectations."

      And the same is true of every "ARM server" Powerpoint I have ever seen.

      1. Mike Pellatt

        Re: Intel Broadwell-D

        Just as Intel have been unable to break into the low-power/embedded market to date (and into the server market with reduced-power x86), ARM have been unable to break into the server market.

        Will either of those change ?? Wish I knew. One or the other might happen at some point. The trick is recognising that point.

        The transputer was gonna wipe the floor with the 80386....

      2. Charlie Clark Silver badge

        Re: Intel Broadwell-D

        And the same is true of every <strikethrough>"ARM server"</strikethrough> Powerpoint I have ever seen.

        Fixed it for you: the same could be said for Intel breaking into the mobile market. These things should be decided in the marketplace, assuming vendors are prevented from anti-competitive behaviour.

        From the Anandtech article:

        The 40nm X-Gene can compete with the 22nm Atom C2000 performance wise, and that is definitely an accomplishment on its own. But the 40nm process technology and the current "untuned" state of ARMv8 software does not allow it to compete in performance/watt.

        Pretty stupid to compare 40nm geometries to 22nm ones as the article makes quite clear.

        Nevertheless, according to Andreas Stiller at Heise, the CERN team reckons the X-Gene is getting close to Xeon:

        CERN-Wissenschaftler haben allerdings vor ein paar Monaten mit dem hauseigenen ParFullCMS-Benchmark etwas bessere Er- gebnisse mit dem X-Gene 1 erzielt, jedenfalls im Vergleich zum nicht so energieoptimier- ten Xeon E5-2650.

        So, the real test will be on standardised architecture with similar geometries.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like