Discussion in 'Computer Information' started by Andy Petro, Dec 4, 2011.

  1. Andy Petro

    Andy Petro Guest

    When I use Hibernate my computer's hard drive keeps running all the time.
    Is this normal , Does it shorten the life of the drive when it never stops
    Andy Petro, Dec 4, 2011
    1. Advertisements

  2. Andy Petro

    Paul Guest

    S1 = Standby (blanks the monitor, all fans running, all drives running, CPU running)
    S3 = Suspend to RAM (ram runs from +5VSB, all fans off, all drives off, CPU off)
    S4 = Hibernate (session stored on disk, all fans off, all drives off, CPU off)

    You're in Standby. Or at least your symptoms *sound* like standby.

    No, it's not normal to select Hibernate, and be in a Standby S1-like state.

    On my WinXP computer, selecting shutdown and holding the "Shift" key, changes
    Standby to Hibernate in the dialog choices. If you didn't hold down Shift,
    you'd see a Standby, and depending on if the OS was screwed up, the
    response to Standby could be S1 or S3. That's a "normal screwup", which
    can be fixed by the "dumppo" utility from Microsoft. If you select Standby
    expecting S3 Suspend to RAM, and all fans remain running, dumppo can
    change S1 to S3 for the OS. Then the fans will stop running when you
    select Standby from the shutdown dialog.

    Selecting Hibernate from that dialog, should have no alternatives. It
    shouldn't vary between S4 and other states. What could happen though,
    is you select Hibernate, session is stored to disk, but because the
    PS_ON# signal isn't working properly, the power supply won't turn off.
    This leaves the computer in a zombie state, with the fans running. You might
    also have noticed problems turning the supply on from a cold start. Like,
    if the back switch was off for a while, you turn on the back switch,
    but pressing the front power button, nothing happens. The PS_ON# signal,
    which flows over the main power bundle (the 20 or 24 pin connector on
    the motherboard), is what controls the enabling or disabling of the
    main section of the power supply. When that happens, it can be a
    motherboard or a power supply issue. I'm baffled by why the motherboard
    driver for that signal, fails as much as it does, but it happens.

    So if you select Hibernate, and the fans remain running, it could be
    a PS_ON# logic signal failure.


    The ATX power supply has two halves to it

    1) +5VSB supply, doesn't need ATX fan to remain cool.
    Also, does not power case cooling fans. It powers the RAM
    during Suspend to RAM.

    2) Main supply half. 3.3V, 5V, 12V, -12V. 12V runs CPU and all the fans

    When the computer remains running, it means the main supply is still on.
    When in Suspend to RAM, the main supply is supposed to be off and
    the +5VSB supply remains on.

    To turn off +5VSB completely, you use the switch on the back of the computer.

    Using the above table of values -

    S1 = Both halves of supply still running. All fans running. Only video output
    to monitor has faded to black.

    S3 = +5VSB half of supply still running. Session will be lost if you turn off
    the computer at the back.

    S4 = +5VSB half of supply still running. Since session is stored on disk, you can
    turn off the computer at the back. RAM contents will be ignored.

    The table here, shows a few more of the possible ACPI values.


    (S5) Soft Off - Same as S4, only session is not stored on disk. Windows does a
    fresh boot when you push the front button. This would happen if
    you select "Turn Off" in the shutdown dialog.

    G3 Mechanical Off - That's the switch on the back. Both halves of supply are now off.

    When you select "Turn Off" in the shutdown dialog, does the computer
    actually soft off and fans stop spinning ? If it did, that would prove
    PS_ON# signal works. And then the question would be, what state are you
    actually in, when you select Hibernate ? Is it a zombie state ? And
    if so, why can it shut off when "Turn Off" is selected and not
    when "Hibernate" is selected ? Both would use the PS_ON# signal.

    Paul, Dec 4, 2011
    1. Advertisements

  3. Andy Petro

    Chas Guest

    Can one access the classes of Standby...S1, S3, S4 without juggling the
    switch on the rear of the cabinet?

    .. . . .Especially using S3 from the keyboard, command, or screen.

    Lastly, how difficult is it to recover from a S3 standby?
    Turning off my monitor and/or wireless mouse is no problem..just shutting
    down the Hdrives and fans and yet keeping the motherboard cool is what I am
    unaware how to do..
    .. . .thank you. . . Chas
    Chas, Dec 5, 2011
  4. Andy Petro

    Paul Guest

    S1, S3, S4 are ACPI (software category) states. The OS and the BIOS, use
    ACPI as a replacement for APM. The OS and the BIOS, own the computer at different
    times of its operation, and there has to be some communication and coordination
    between them (operational handoff, from one to the other). If you look in
    the BIOS, you may even see some settings for it.

    If your OS lacked ACPI support, you'd see the old "it is now safe to turn
    off your PC" screen, after shutdown. In that case, the OS lacks the
    ability to tell the BIOS to turn off the power. The user then flips the
    switch or pushes the button themselves, manually. The same thing can happen,
    if the "Computer" entry in Device Manager, doesn't include the word "ACPI"
    in the HAL type. (Such a thing can happen, if the install is done wrong,
    or if the BIOS suffers some issues where it no longer passes the
    appropriate ACPI tables to the OS.)

    "ACPI Multiprocessor PC" <--- mine, right now - says it supports ACPI

    If your Computer entry in Device Manager is correct, it tells you the
    OS is currently running in a way that supports ACPI, and that includes
    having detected ACPI support from the BIOS. At that point, with a
    "good" HAL entry in Device Manager, you can use "dumppo" to detect
    whether the OS supports S1, S3, S4 and so on. There is a little
    more info on dumppo, here.


    The BIOS may have an ACPI setting, "S!", "S3", "S1&S3", and
    sometimes that needs to be adjusted, before doing an administrative
    override with "dumppo". That is useful for correcting a system
    that is going into S1, instead of the quieter S3 state. I don't
    know if there is anything useful you can do with "dumppo" though,
    to fix a Hibernate S4 problem. "Dumppo" is typically used to
    convert an S1 system ("fans still running") to S3 ("fans off").
    In S3, the OS current session, is still stored in the RAM modules
    and +5VSB is still running, to keep the RAM contents. That uses
    maybe 10 watts of power from the outlet, and the drives and
    fans won't be spinning in S3.


    Physically, there are two ways to turn off the power.

    In the Southbridge, there will be a register in there somewhere,
    with the ability to control the PS_ON# signal (or logic in that
    path). You could probably write a program, and run it under plain
    MSDOS (boot from a DOS floppy), to access and trigger that register.

    But modern OSes, have a degree of isolation between user space
    and the kernel. And the deal with the kernel, is it's agreed to
    use the ACPI specification. Which means, it asks the BIOS politely,
    to go access the register and flip the bit for it. (They used
    to need a lot of custom code, to do this in the past, which
    is why they decided to have a division of labor, such that
    Microsoft doesn't have to do anything when new motherboards
    are released. Adherence to ACPI, means the work is done by
    the motherboard manufacturer.)

    So the Southbridge register, is the way it can be controlled.
    But getting there, is more involved in an ACPI environment.
    More layers of "handshaking and saluting" before it gets done.

    The "mechanical" switch in the back, is certainly easily accessible,
    and also gets the job done. If done at the wrong time, your
    file systems can be corrupted and need CHKDSK at startup. You
    use the mechanical switch in an emergency, such as if the
    PC is on fire. But otherwise, try to coax ACPI and the BIOS,
    to do the right thing, via the PS_ON# "soft power" control.
    That's all part of a more orderly shutdown, where the
    file system state is flushed to disk, before the power
    goes off.

    Paul, Dec 5, 2011
  5. Andy Petro

    Chas Guest

    I'm dying to go into your suggestions but right now I suffer from the
    syndrom 'If it's not broke, don't fix it'.
    .. . .Just got this machine up and running a few days ago, Had some setbacks
    but now it is tremendous. I had a BIOS battery (6 yrs old) I suspected was
    weak and I replaced it. That was not the culprit...I have a single EIDE
    cable port on my motherboard and that port was losing its settings between
    boots. I have a SAT Hdrive I boot from so that was no problem. Also a
    Winchester 60 GiG Hdrive that is EIDE and it was sporatic in 'showing up'
    (along with my CD drive). I finally changed the 'interupts' on the EIDE
    port and all now seems to be great...
    .. . .I put an amp meter on my 120v power supply line and determined it cost
    me $8 / mo ($40 on the East coast) to shut off everything but the
    motherboard. But the recovery time was pretty long to get the drives up and
    spinning again.
    *******Incidentally, the lowest amp reading (with a 10X multiplier) that I
    was able to observe was 1.1 amps.. . . .times 120v = 132 watt/hrs.. . .X24
    hrs=3168 watt/hrs/day. . . .X31 days=9808 watt/hrs/mo . . .divided by 1000
    kiloW=98.2 KW/mo. . .
    X $0.08 KW/hr =$7.86.... hope I haven't messed this up..<G>.
    . . . thank you for your help...charlie
    Chas, Dec 6, 2011
  6. Andy Petro

    Paul Guest

    The best way to do the power measurement, is with a Kill-O-Watt meter.

    The thing is, the ATX supply has a power factor, and the current flow
    waveform is not a perfect sinusoid. You'd need a "true RMS" meter, to
    properly measure the waveform. True RMS meters have a bandwidth limit,
    like only being able to measure and include harmonics up to perhaps 50KHz
    or so. You can see the current waveform, has a rather steep edge
    in this picture. Probably doesn't go out to 50KHz, but it's still
    going to tax the abilities of the true RMS meter.


    The Kill-O-Watt meter, will measure both the voltage and current at
    the same time, and do the math to determine the real and imaginary
    components of power, then only display the real (billable) component.
    And that's what makes that kind of meter more useful than an ammeter.
    The Kill-O-Watt meter uses the same methods that the digital meter
    outside your house might use.

    I too, have tried what you've done, but the results bothered me.
    I was getting silly results for an unloaded ATX supply, so I've more
    or less stopped using that method (measuring just the current,
    assume a particular value of power factor, using the current line
    voltage measured value, and working out power from that).

    I don't own a Kill-O-Watt meter. They're relatively cheap. If I
    saw one at a local retailer, I'd probably buy one. So far, I've only
    seen them for sale on the Internet, and there, shipping or border
    brokerage fees would take the fun out of it.


    Paul, Dec 6, 2011
    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.