back to article HP test pilots fly secret four-socket Xeon E5 blade in public

Hewlett-Packard has yet to announce a blade server based on Intel's latest Xeon E5-4600 processors for four-socket servers - but that hasn't stopped it bragging about their performance. In an article on its website, HP let the cat out of the bag that it was working on the ProLiant BL660c Gen8 blade server, which is a full- …

COMMENTS

This topic is closed for new posts.
  1. PracticalApplications
    FAIL

    These tests are (of course) bogus

    EX & EP is a ridiculous comparison. Cache structures, cores, just about everything is different and integer sensitive workloads will fly on EP and do worse on EX. Where's the compare to the other 4 Socket E5-4600 EP's? Like the IBM x440 (http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=AN&subtype=CA&htmlfid=897/ENUS112-099&appname=USN). So the net result is that HP is using an exceptionally defective method of comparison. Huh... where's the Itanium compares? <crickets>

    1. Matt Bryant Silver badge
      Happy

      Re: These tests are (of course) bogus

      ".....Like the IBM x440...." Surely Armonk is much more worried about the comparison of any of the new Xeon blades to the over-priced and under-performing Power blades?

      1. Anonymous Coward
        Anonymous Coward

        Re: These tests are (of course) bogus

        I suspect power doesn't worry about xeon in the same way xeon doesn't worry about core i7.

        As for HP, surely in a world of identical CPUs, you add infrastructure - management, fault tolerance etc.

      2. Anonymous Coward
        Anonymous Coward

        Re: These tests are (of course) bogus

        IBM has posted results for the x440 here:

        http://public.dhe.ibm.com/eserver/benchmarks/news/newsblurb_x440_speccpu_20120807.pdf

        Believe it or not....the identical processors produce identical results....imagine that..

        One day I would really like to see some meaningful benchmarks like average time to repair, or help center hold time.

        1. theblackhand

          Re: These tests are (of course) bogus

          Re: service benchmarks

          "And you will notice that the Dell models fair badly at this step where the help desk person becomes stuck in the reboot loop, repeatedly asking the customer to reboot the server in the hope that it will somehow fix the issue...."

    2. Anonymous Coward
      Anonymous Coward

      Re: These tests are (of course) bogus

      True, you cannot compare EX to EP. meaningless benchmark.

  2. Anonymous Coward
    Anonymous Coward

    The data is incorrect. For example, Supermicro platform has specint numbers as 1220 in spec.org. http://www.spec.org/cpu2006/results/res2012q2/cpu2006-20120507-21657.pdf

This topic is closed for new posts.

Other stories you might like