No announcement yet.

Changes to original RAM config causes severe performance hit

  • Filter
  • Time
  • Show
Clear All
new posts

  • Changes to original RAM config causes severe performance hit

    Hi -

    I inherited some moderately older hardware, several identical HP dc7700 machines. Each came with 1GB RAM configured as 2x512 MB PC2-5300 sticks.

    I ran Passmark tests on a few of them and they pegged the values I expected from my research.

    I added another 2x512 of RAM to one, which was cannibalized from one of the other identical machines. All four sticks were the same model, part number, and even from the same batch/lot. The machine seemed to run slower, so I ran the tests again and, sure enough, it benchmarked at about one-fourth the original value, with the memory tests suffering the most.

    I concluded that one of the added sticks must be bad, so I went back to the original config and benchmarked again and got my original results.

    Then I tried removing one stick, leaving the machine with just 512 MB. When I ran the tests, they were just as slow as when I ran it with all four, which makes me think the issue is not that one of the additional sticks was bad.

    Does anyone have any thoughts on why this might be? I've read that dual-channel operation can have a performance benefit, by what I've read suggests maybe on the order of 10%... not 400%. Even if that were the case, it still wouldn't explain the poor performance with 4x512.

    Interested in any feedback. Thanks in advance for your help!
    Last edited by bkann; Dec-04-2012, 06:01 PM.

  • #2
    I'm going to go ahead and plead stupidity here. The machines' BIOSes were several versions behind and updating them seems to have fixed the problem. In fact, the issue was specifically identified in one of those versions. Sorry for the premature post and thanks for looking.


    • #3
      Glad the our software was able to help you identify the problem.