1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

Is 250/200 better than 220/220?

Discussion in 'Overclocking' started by Tony, Jan 18, 2004.

  1. Tony

    Tony Guest

    Or is the SiSandra memory benchmark really not the best way to check memory
    performance? I have been doing some testing (and oc'ing) on my Abit AI7, on
    which I am running 2 512MB sticks of Kingston HyperX PC3500. I had been
    running 220MHz 1:1, which the SiSandra memory benchmark produced a result of
    5000 MB/s Buffered. I decided to see how far I could push the CPU (P4 2.6C).
    I got to 260MHz with some stability problems (I probably could have
    corrected that, but did not want to push the CPU voltage beyond 1.65 volts).
    So I backed off to 250MHz at 5:4. Memory running at 200MHz. The memory
    benchmark produced a result of 5350 MB/s! Is SiSandra being overly
    influenced by the CPU in their memory benchmark? Is this really an accurate
    indicator of memory performance? Is it better to run at 250/200 than
    220/220?
     
    Tony, Jan 18, 2004
    #1
    1. Advertisements

  2. Tony

    ICee Guest

    You may want to try www.memtest86.com . It runs in DOS from a floppy
    and reports your memory bandwidth (as well as testing memory). This
    should give you an accurate measure of which is better.
     
    ICee, Jan 18, 2004
    #2
    1. Advertisements

  3. Tony

    Tony Guest

    Yes, I have memtest86. I've noticed the bandwidth readout but wasn't sure if
    that was a better (more realistic) assessment of memory performance. I'll
    give it a try.
     
    Tony, Jan 18, 2004
    #3
  4. Tony

    ICee Guest

    Not sure if it's accurate, per se, but it will show without bias which
    setting has the higher bandwidth.
     
    ICee, Jan 18, 2004
    #4
  5. Tony

    Skid Guest

    Are your memory timings the same at 200 and 220? If the timings are set by
    SPD, it could be that they change from, say CAS 2 at 200 to CAS 2.5 (or 3)
    at 220. If so, that would explain your results. Sandra is notoriously
    fickle, and results can vary with the phase of the moon and which side of
    the bed you got out of that morning.

    In pure memory bandwidth, 220 should be faster if all other things are
    equal. However, in overall system performance, your 2.6 will run a lot
    better at 3.25 ghz than at 2.86. If the bandwidth is also faster, it's a
    no-brainer to run 5:4, 250/200.
     
    Skid, Jan 18, 2004
    #5
  6. Tony

    Tony Guest

    Yes. I have it set manually to 2-7-3-3, which works at 220MHz. I just left
    it that way for the 250/200 test.
     
    Tony, Jan 18, 2004
    #6
  7. Tony

    Tony Guest

    Well, you guys were on to something. memtest86 showed a memory bandwidth of
    2727 at 220MHz and only 2447 at 200MHz. Makes sense when looked at that way,
    but not in keeping with the SiSandra results. The question is: which is a
    more accurate reflection of performance in Win XP?
     
    Tony, Jan 19, 2004
    #7
  8. What you really want to know, I presume, is which combination of CPU speed
    and memory bandwidth is 'better' in WinXP.

    Encode a video, like convert an AVI to MPG, at both settings and see which
    is faster.
     
    David Maynard, Jan 19, 2004
    #8
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.