Announcement

Collapse
No announcement yet.

Calxeda EnergyCore ECX-1000 ARM Server

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Calxeda EnergyCore ECX-1000 ARM Server

    Phoronix: Calxeda EnergyCore ECX-1000 ARM Server

    Earlier this month I was down in Texas visiting the Calxeda office where for the past four years they have been busy trying to revolutionize the server market through ultra-low power ARM-based servers. This morning one of their partners, Boston Limited, is formally launching their energy-efficient "Viridis" server built around Calxeda's EnergyCore ECX-1000 hardware. In this article are the first of some public Calxeda ARM benchmarks on Ubuntu Linux while more results will be out in the coming days.

    http://www.phoronix.com/vr.php?view=18011

  • #2
    Some new hardware comes out, and the first thing you compare it to... is itself?

    Comment


    • #3
      Very cool machine! Thanks for review! True, needs something to compare to, however... probably an Intel or AMD equivalent, running in same total cost segment.

      Comment


      • #4
        Per-watt numbers when?

        Comment


        • #5
          Michael, maybe you should try to show the phoronix test suite to these people: http://phoronix.com/forums/register....f931f51f7b2d6c

          Do a proper benchmark, taking into account all the points made in the article.

          Also, any idea about prices? Of single chips and complete minimal setups, like what's offered by HP?

          Comment


          • #6
            The best thing to run this against would be something like Atom or Tegra 3. For comparison you could also calculate some performance per watt metrics for Atom, Tegra 3, and Ivy Bridge. That'd be cool.

            I bet the performance per watt on some of the old dedi servers people run (Core 2 Quad, etc) is heinous! I'd love to see performance per watt on that "venerable" old hardware tested against something cutting edge like this.

            Comment


            • #7
              Originally posted by allquixotic View Post
              The best thing to run this against would be something like Atom or Tegra 3. For comparison you could also calculate some performance per watt metrics for Atom, Tegra 3, and Ivy Bridge. That'd be cool.

              I bet the performance per watt on some of the old dedi servers people run (Core 2 Quad, etc) is heinous! I'd love to see performance per watt on that "venerable" old hardware tested against something cutting edge like this.
              Intel Atom, AMD Bobcat, and maybe a low end ivy bridge i3 would make a good comparison.

              Comment


              • #8
                This looks like an interesting chip - although the 32-bitness isn' ideal.

                I expect a future ARMv8 variant of this device built on a 20nm node will be very competitive.

                But yeah, it would be nice to see it compared with comparable Atoms and Bobcats.

                Comment


                • #9
                  Comparing the results to http://openbenchmarking.org/result/1...RA-INTELATOM04, it looks like this quad-core ARM at 1.4 GHz can keep up with an Intel Atom D525 at 1.8 GHz. So the flagship quad-core ARM product ist equivalent to a dual-core Atom from 2010, the Atom running at a slightly higher clock frequency.

                  Comment


                  • #10
                    32bit isn't all that big of a hit if you are using this for webhosting... if you were using it for serving out databases and such it probably would be since those are going to be geard toward big iron thats already 64bit an example being ZFS

                    Comment


                    • #11
                      Originally posted by sturmflut View Post
                      Comparing the results to http://openbenchmarking.org/result/1...RA-INTELATOM04, it looks like this quad-core ARM at 1.4 GHz can keep up with an Intel Atom D525 at 1.8 GHz. So the flagship quad-core ARM product ist equivalent to a dual-core Atom from 2010, the Atom running at a slightly higher clock frequency.
                      Comparing a low-end/low-power desktop board that has a GPU, WiFi, etc. with this low-power server board which includes support for things like ECC memory and built-in IPMI probably isn't very useful...

                      What you really want to benchmark is how this ARM system (and especially a number of them working together) fares doing several typical server tasks, and compare that to other systems/clusters doing the same. And take TCO, including space savings & power consumption, into account while doing so.

                      Because clustering a whole bunch of these to work together is what they are supposedly good at:
                      http://www.calxeda.com/technology/architecture/fabric/
                      That's an on-chip network switch with 8 ports (3 internal, 5 external) that can handle a total capacity of up to 80 Gbit/s, or up to 10 Gbit/s per channel, which can also automatically slow down (to save energy) depending on the current requirements. To compare: a standalone ethernet switch which can handle speeds like that would consume more than 5W on its own (and also require a lot more space, of course).

                      Comment


                      • #12
                        Originally posted by cb88 View Post
                        32bit isn't all that big of a hit if you are using this for webhosting... if you were using it for serving out databases and such it probably would be since those are going to be geard toward big iron thats already 64bit an example being ZFS
                        Depends on what type of database...

                        You might be right when you talk about a big SQL database that because of its structure doesn't lend itself well to sharding or similar techniques. OTOH, databases like CouchDB and Hadoop might do pretty well, spreading over more nodes as the data grows.

                        Comment

                        Working...
                        X