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.

Can MEMTEST86+ check memory running at faster bus speed?

Discussion in 'AMD Overclocking' started by Franklin, Oct 14, 2004.

  1. Franklin

    Franklin Guest

    I have set my motherboard/BIOS to exceed the normal the bus frequency
    by just a little bit. Now I would like to test my memory to see if
    it is ok under the new settings.

    Memtest86+ from http://www.memtest.org/ looks like a good memory
    tester. You make a special floppy and boot from the floppy. So it
    tests before Windows is launched.

    Does Memtest86+ test memory taking into account my newly chosen bus
    frequency settings?

    Or does it somehow ignore most motherboard settings (like my
    frequency increase) and test the memory "raw"?







    [ crossposted. relevant groups]
     
    Franklin, Oct 14, 2004
    #1
    1. Advertisements

  2. Franklin

    Shep© Guest

    No software RAM testers are that much use IMHO.Memtest86+ won't point
    out anything of relevance in overclocking,none of them will.
    My 0.2




    --
    Free Windows/PC help,
    http://www.geocities.com/sheppola/trouble.html
    remove obvious to reply
    email
    Free songs to download and,"BURN" :O)
    http://www.soundclick.com/bands/8/nomessiahsmusic.htm
     
    Shep©, Oct 14, 2004
    #2
    1. Advertisements

  3. The BIOS controls the memory timing, the memory will run at whatever
    speed you've set it at. Memtest86 is as good a test as ay to see if your
    memory still works. Be advised that by overclocking your system you've
    given up timing margins. Just because it works when the room is cool or
    when you system is under light load doesn't mean it will work if the room
    gets 10 degrees warmer.
     
    General Schvantzkoph, Oct 14, 2004
    #3
  4. Franklin

    kony Guest

    There is no such thing as "raw" testing that isn't dependant
    on the host device setting the memory bus speed. In other
    words, yes it tests "overclocked" settings, which to a
    memory module means just another (any particular) speed,
    it's all relative.
     
    kony, Oct 14, 2004
    #4
  5. True that software testers are limited in their capabilities but to say
    that none will "point out anything of relevance" is absurd and
    ill-informed. Have you even tried it?

    Memtest86+ is actually a good, if not the best, software based memory
    tester. It is certainly a very good confidence check that nothing is
    horribly awry and I consider it standard practice to run Memtest86+ for a
    couple of hours before attempting installation of an OS. IME, a system
    which has passed the checks - as well as a hard disk diagnostic - has
    always installed and run the OS without problems.

    Rgds, George Macdonald

    "Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
     
    George Macdonald, Oct 15, 2004
    #5
  6. Yes, chipset/bus freq is set by the BIOS when the machine
    boots, not by the OS.

    memtest86 is a very good, extensive, memory tester.
    It is not a intensive (high bandwidth) as I would like,
    so I wrote some in my CPUburn package. Try `burnMMX`.

    -- Robert author `cpuburn` http://pages.sbcglobal.net/redelm
     
    Robert Redelmeier, Oct 15, 2004
    #6
  7. Franklin

    CrackerJack Guest

    Looks neat.

    Does your CPUburn have any special points when compared to the cpu
    testers discussed at Radifed?

    Like Prime95, Motherboard Monitor's 'Heat Up', HotCPU Tester Pro
    Lite, etc.

    http://radified.com/Articles/stability_testing.htm
     
    CrackerJack, Oct 15, 2004
    #7
  8. Franklin

    Mixxy Guest



    Has anyone tried Metabench? It is in late beta development and is still
    free.

    http://www.7byte.com/index.php?page=metabench

    Is it any good?
     
    Mixxy, Oct 15, 2004
    #8
  9. Franklin

    spodosaurus Guest

    Yes. Memtest86 performs a series of read writes to the memory and
    doesn't bugger about with clock speed as a 'problem detection' feature.
    If all the read/writes work, then it gets a pass, if not errors are
    reported. It doesn't say 'try lowering your clock speed and see if these
    errors go away'. You'll have to figure that yourself :)

    --
    spammage trappage: replace fishies_ with yahoo

    I'm going to die rather sooner than I'd like. I tried to protect my
    neighbours from crime, and became the victim of it. To jump to the end
    of the story, as a result of this I need a bone marrow transplant. Many
    people around the world are waiting for a marrow transplant, too. Please
    volunteer to be a marrow donor:
    http://www.abmdr.org.au/
    http://www.marrow.org/
     
    spodosaurus, Oct 15, 2004
    #9
  10. I really haven't had much time to look around. If I had the
    time, I'd be releasing `burnRAM` [need win32 port] and `burnP7`
    [needs some signals work].

    It's very easy to get "100% CPU utilization" according to
    the OS. `jmp $` or `while(1);` will do. The OS always has
    something to run (not the idle thread), so it thinks it's busy.
    If you can't get 100% (MS-Win9*), it's a priority issue.

    But this is only around 70% of max power draw. Not all the
    chip circuits are kept busy. I've crafted my burn* pgms in
    assembly (natch!) to try to keep as much busy as possible.
    Without any constraint of actually doing useful work!

    Some programs can keep the CPU 100% runnable but really not
    be compute-limited. Doing useful work is a bit of a limit.
    I stuff useless instructions in. [email protected] was notorious for
    very odd times for work unit completion (memory fetch bound).

    I _don't_ claim my pgms are the hottest possible. I'm sure
    that Intel and AMD use better ones as part of their CPU
    manufacturing testing. But those are deep dark secrets.
    Mine is Open Source.

    -- Robert author `cpuburn` http://pages.sbcglobal.net/redelm

    (email invalid, changed ISP -- you figure it out)
     
    Robert Redelmeier, Oct 15, 2004
    #10
  11. Franklin

    S.Heenan Guest


    Press "c" "2" "3" "Enter" to run all eleven tests.
     
    S.Heenan, Oct 15, 2004
    #11
  12. Franklin

    Shep© Guest

    Yes.Next to useless and not worth the download let alone running.

    Window's itself is a good test of hardware memory.It will balk if
    there's anything wrong usually throwing up a,"Registry" fault.
    Why some people defend a piece of software that they get for free I'll
    never know.Hardware RAM testing machines run into the thousands.Go
    figure.
    PS
    Best RAM test if the user suspects a fault is to swap with a known
    good stick ;-)




    --
    Free Windows/PC help,
    http://www.geocities.com/sheppola/trouble.html
    remove obvious to reply
    email
    Free songs to download and,"BURN" :O)
    http://www.soundclick.com/bands/8/nomessiahsmusic.htm
     
    Shep©, Oct 15, 2004
    #12
  13. Franklin

    Rich Webb Guest

    Oddly enough I seem to recall a Shep from the K7S5A boards, where
    Memtest86 was the recommended tool to identify motherboards that had
    "issues" with the higher speed Athlon T-birds. Maybe a different guy...
     
    Rich Webb, Oct 15, 2004
    #13
  14. Defective software, drivers, and other MS-Windows
    cruft can also throw lockups, BSoD & Reg.errs

    When these happen, you don't know if it's hardware
    or software. Best to have some simple testers that
    can rule out hardware. Testers can usually be more
    intense than apps or OSes.

    -- Robert
     
    Robert Redelmeier, Oct 15, 2004
    #14
  15. Franklin

    assaarpa Guest

    Yes.Next to useless and not worth the download let alone running.

    I had unstable system. I ran a Linux based memory tester which booted off a
    cdrw or 3.5" disk, and guess what? One of the Dual Channel modules was
    malfunctioning, I went to the shop (computeria.fi, mind you) and got
    replacement stick no questions asked. It worked flawlessly. Solved the
    problem for me.
    Yeah. So good. The system was unstable with defective ram, and guess what?
    The machine just froze completely randomly, after 2 minutes, 7 minutes.. but
    only if using a specific graphics intensive application. Initial reaction
    was that maybe the driver is faulty, since that is not unfamiliar thing to
    me after using ATI and nVidia products for years. So I switch driver. I
    switch card. I switch vendor. Still keeps crashing. For some reason after
    hours of tinkering, I somehow just know it is the ram, I download the
    tester, burn it to cdrw.. boot with hands trembling.. and what the hell,
    defective ram! After getting the sticks (dual channel kit!) replaced,
    everything works like charm and has ever since (posting from that very same
    system, A64 3000+ K8V DLX).

    Latest problem was SP2 upgrade, Windows XP kept bluescreening but MS KB had
    article about that, apparently the DEP / NX was broken in Windows XP SP2..
    now SP2 works well too thanks for asking. I wonder what breaks next. ;-)
    Good idea. Next time I buy a new computer I will buy two dual channel 1GB
    memory kits, just in case. No damn, what am I saying.. I didn't listen to
    you at all, what I should do is to buy a spare "known good stick", how the
    hell I know a stick is a known good stick anyway until I test it? I would
    ASSUME that when I pay hundreds of bucks for a known Brand Name stick the
    manufacturer would have somekind of quality assurance and testing procedure,
    right? I don't know if they do, but sure as hell a defective sticks slipped
    through.

    So.. how you propose we know what stick is a good one and what isn't? Oh, by
    testing? A marvelous idea.. howcome we didn't think of that.. *slaps
    forehead* ..
     
    assaarpa, Oct 15, 2004
    #15
  16. A registry fault is a memory problem?<guffaw> There's a helluva lot of
    other things in both software and hardware which can cause instability in
    Windows... or any other OS for that matter.
    Memory testing software is nothing new - it was used on mainframes and
    minicomputers for years. It can generated crafted memory access patterns
    which may occur once a day or less in a running OS.
    With a memory tester, even a software one, you'll be closer to *knowing*
    that it's the memory. A couple of hours of intensive memory testing,
    *before* loading the OS, can save you a lot of grief and time.

    Rgds, George Macdonald

    "Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
     
    George Macdonald, Oct 16, 2004
    #16
  17. Franklin

    CrackerJack Guest

    ISTR there are now 12 tests. Right?
     
    CrackerJack, Oct 16, 2004
    #17
  18. Franklin

    S.Heenan Guest



    That may well be the case. I can not remember trying extended tests in the
    newest version.
    I imagine the same keystrokes apply.
    http://www.memtest.org/
     
    S.Heenan, Oct 16, 2004
    #18
  19. Franklin

    Franc Zabkar Guest

    I agree, but I recall one particularly troublesome memory board in a
    minicomputer during the 80's which was not faulted by regular
    diagnostic software. This software was very intensive, probably more
    so than Memtest-86. It generated many different patterns, and tested
    for interference between adjacent memory cells. I ran this software
    for several days but was not able to fault the board. However, the OS
    and/or application software would crash about once a day with a parity
    error. The OS was able to trap the address of this error, but could
    not identify the faulty bit. As each bit was stored in a different
    DRAM chip, I was facing the prospect of desoldering and replacing up
    to 17 chips (16 + parity). Fortunately I eventually narrowed down the
    faulty bit after writing a very simple diagnostic routine to exercise
    this one location in a tight loop.

    The one reservation I have with RAM testing software is that some
    appear to have no adequate test for refresh problems. Faulty refresh
    logic is more likely to show up in normal use, but not during a memory
    test when cells are accessed (and therefore refreshed) continuously.


    - Franc Zabkar
     
    Franc Zabkar, Oct 17, 2004
    #19
  20. I can recount a few stories about diagnostic software that missed a
    particular type of fault too but that doesn't mean they were useless.
     
    David Maynard, Oct 17, 2004
    #20
    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.