• Jeff Moden (3/18/2008)


    And then you better make sure you run the code more than once, even on a quiet system...

    I did three runs in one order, and two runs with the order switched, the % cost matched, though the % time did vary slightly. The issue that threw me off was that an I/O intensive load was running at the same time, so the I/O portion of the costs was actually taking much more time relative to the CPU and RAM portions.

    Then again, it may just be that the cost formula is not properly balanced for our server for some reason that isn't apparent to me.


    Puto me cogitare, ergo puto me esse.
    I think that I think, therefore I think that I am.