Re: Stream results for dual PIII and dual Athlon

From: James Cook (cook@ned1.sims.nrc.ca)
Date: Mon Aug 19 2002 - 09:37:37 CDT

  • Next message: Mikhail Kuzminsky: "new STREAM results"

    On Fri, 16 Aug 2002, John D McCalpin wrote:

    > These results are, unfortunately, typical of IA32 systems.
    > Intel has designed the processor to be able to saturate the
    > front-side bus, so adding another CPU provides no benefit.
    > In some cases there is a minor degradation from the extra
    > address traffic.
    >

    Yes. I discussed the results with some people and they said the same
    thing. Thanks. However, I have some MPI results and I got a bit of a
    performance increase running on 2 cpus for the athlon yet no increase in
    performance for the PentiumIII.

    We have a 16 node cluster using the same computers I did the previous
    tests on. Of the 16 computers, 9 are Pentium-III@933Mhz. The other 7 are
    the Athlon 1900+ MP at 1.6Ghz. Each have a Dolphin D339 SCI card using
    Scali MPI (which comes from Scali SSP 3.0.1)

    I have attached the results. I ran the following tests:
    1 Pentium, 1 cpu
    1 Pentium, 2 cpu
    1 Athlon, 1cpu
    1 Athlon, 2cpu
    4 Petium, 2cpu
    4 Athlon, 2cpu
    And all 16 nodes.

    Enjoy!

    James Cook
    Steacie Institute for Molecular Sciences
    National Research Council Canada

















    This archive was generated by hypermail 2.1.4 : Fri Nov 08 2002 - 13:37:15 CST