RetroBBS - alt.comp.hardware.pc-homebuilt - New problem with starting PC from sleep mode

Bob F wrote:
> I have a Win 8.1 PC that has been working fine for months but now will
> not restart from sleep properly.
>
> When I push the power button to wake it, the power light comes on
> (blue), then it just hangs there and never brings up Win 8.1. The disk
> LED does not flash at all.
>
> If I hold the power button until it turns off and then push it again, or
> just push the reset button, then it starts up just like it used to and
> brings up the previous windows state. The disk LED flashes, and the
> power LED changes to yellow.
>
> It used to restart very quickly from sleep operating from the SSD drive.
>
> I have tried:
> Resetting the BIOS to defaults.
> Turning off the fast boot function in BIOS - this one causes it to
> display the BIOS splash screen during the hang state.
> Setting the network card to turn off when the computer is turned off.
> Disabling the "fast restart" in the power options page.
>
> Can anyone offer suggestions to solve this problem?

You could test with a second OS that supports sleep,
to see if the problem is reproducible (a way to prove
a hardware defect).

You can also try reducing the number of USB toys
plugged into the PC. Stop charging your iPhone
off the back, as an example of a naughty thing to do.

The +5VSB that powers the RAM sticks during sleep,
it can't handle more than 2.5A to 3A loading
(check the PSU label for details on +5VSB).

This power is reserved for PC usage, and is generally
sufficient in most situations.

However, plugging in a 2 ampere "Apple Load", that
is pushing your luck. The PC could easily be using
1 ampere to 1.5 amperes for various functions when
the PC is sleeping. An "Apple Load" could push it
over the edge, causing the PSU to temporarily shut
off completely. Then, RAM content is lost.

While you could try setting up Hybrid Sleep (combines
aspects of S3 and S4, by the S3 also writing out
the Hiberfile just in case), that may not alleviate
the BIOS level problem of "not liking what it sees".

And that's why I would want the opinion of, say,
a Linux OS where you've gone to the trouble of
getting all the ACPI stuff set up properly. I would
start by Googling for a distro that has out-of-the-box
support for sleep and so on. The last time I looked,
the one I was using wasn't immediately ready to sleep.

Paul

On 9/28/2021 4:29 AM, Paul wrote:
> Bob F wrote:
>> I have a Win 8.1 PC that has been working fine for months but now will
>> not restart from sleep properly.
>>
>> When I push the power button to wake it, the power light comes on
>> (blue), then it just hangs there and never brings up Win 8.1. The disk
>> LED does not flash at all.
>>
>> If I hold the power button until it turns off and then push it again,
>> or just push the reset button, then it starts up just like it used to
>> and brings up the previous windows state. The disk LED flashes, and
>> the power LED changes to yellow.
>>
>> It used to restart very quickly from sleep operating from the SSD drive.
>>
>> I have tried:
>>     Resetting the BIOS to defaults.
>>     Turning off the fast boot function in BIOS - this one causes it to
>> display the BIOS splash screen during the hang state.
>>     Setting the network card to turn off when the computer is turned off.
>>     Disabling the "fast restart" in the power options page.
>>
>> Can anyone offer suggestions to solve this problem?
>
> You could test with a second OS that supports sleep,
> to see if the problem is reproducible (a way to prove
> a hardware defect).
>
> You can also try reducing the number of USB toys
> plugged into the PC. Stop charging your iPhone
> off the back, as an example of a naughty thing to do.
>
> The +5VSB that powers the RAM sticks during sleep,
> it can't handle more than 2.5A to 3A loading
> (check the PSU label for details on +5VSB).
>
> This power is reserved for PC usage, and is generally
> sufficient in most situations.
>
> However, plugging in a 2 ampere "Apple Load", that
> is pushing your luck. The PC could easily be using
> 1 ampere to 1.5 amperes for various functions when
> the PC is sleeping. An "Apple Load" could push it
> over the edge, causing the PSU to temporarily shut
> off completely. Then, RAM content is lost.
>
> While you could try setting up Hybrid Sleep (combines
> aspects of S3 and S4, by the S3 also writing out
> the Hiberfile just in case), that may not alleviate
> the BIOS level problem of "not liking what it sees".
>
> And that's why I would want the opinion of, say,
> a Linux OS where you've gone to the trouble of
> getting all the ACPI stuff set up properly. I would
> start by Googling for a distro that has out-of-the-box
> support for sleep and so on. The last time I looked,
> the one I was using wasn't immediately ready to sleep.
>
>    Paul

This PC is being used in my exercise room as a home theater PC. The only
USB devices are the keyboard, mouse, and an infrared receiver for the
remote control. It does not charge anything. The TV is connected with
with a DVI to HDMI cable. Audio out goes to an audio receiver.

The PC is an old Shuttle micro ATX box with a Q6600 processor and an AMD
HD 6350 video board.

I'm not sure I'm ready to load an entirely different OS to test just
yet. This has been working correctly for a year or 2, and only recently
stopped starting up correctly from sleep.

It seems that the "sleep" memory is working, since it comes out of sleep
properly when I push the reset button after it freezes the first time
after power on. Several seconds after that, the screen is restored
exactly as it was after being put to sleep using the power button. This
has got to be the most significant indicator for the problem. Why does
it not do it correctly when I power it up, but does after a reset.

I tried uninstalling the 3 windows updates installed this month, with no
change from that.

I tried switching to a different power plan and back, and no change from
that.

Bob F wrote:
> On 9/28/2021 4:29 AM, Paul wrote:
>> Bob F wrote:
>>> I have a Win 8.1 PC that has been working fine for months but now
>>> will not restart from sleep properly.
>>>
>>> When I push the power button to wake it, the power light comes on
>>> (blue), then it just hangs there and never brings up Win 8.1. The
>>> disk LED does not flash at all.
>>>
>>> If I hold the power button until it turns off and then push it again,
>>> or just push the reset button, then it starts up just like it used to
>>> and brings up the previous windows state. The disk LED flashes, and
>>> the power LED changes to yellow.
>>>
>>> It used to restart very quickly from sleep operating from the SSD drive.
>>>
>>> I have tried:
>>> Resetting the BIOS to defaults.
>>> Turning off the fast boot function in BIOS - this one causes it
>>> to display the BIOS splash screen during the hang state.
>>> Setting the network card to turn off when the computer is turned
>>> off.
>>> Disabling the "fast restart" in the power options page.
>>>
>>> Can anyone offer suggestions to solve this problem?
>>
>> You could test with a second OS that supports sleep,
>> to see if the problem is reproducible (a way to prove
>> a hardware defect).
>>
>> You can also try reducing the number of USB toys
>> plugged into the PC. Stop charging your iPhone
>> off the back, as an example of a naughty thing to do.
>>
>> The +5VSB that powers the RAM sticks during sleep,
>> it can't handle more than 2.5A to 3A loading
>> (check the PSU label for details on +5VSB).
>>
>> This power is reserved for PC usage, and is generally
>> sufficient in most situations.
>>
>> However, plugging in a 2 ampere "Apple Load", that
>> is pushing your luck. The PC could easily be using
>> 1 ampere to 1.5 amperes for various functions when
>> the PC is sleeping. An "Apple Load" could push it
>> over the edge, causing the PSU to temporarily shut
>> off completely. Then, RAM content is lost.
>>
>> While you could try setting up Hybrid Sleep (combines
>> aspects of S3 and S4, by the S3 also writing out
>> the Hiberfile just in case), that may not alleviate
>> the BIOS level problem of "not liking what it sees".
>>
>> And that's why I would want the opinion of, say,
>> a Linux OS where you've gone to the trouble of
>> getting all the ACPI stuff set up properly. I would
>> start by Googling for a distro that has out-of-the-box
>> support for sleep and so on. The last time I looked,
>> the one I was using wasn't immediately ready to sleep.
>>
>> Paul
>
> This PC is being used in my exercise room as a home theater PC. The only
> USB devices are the keyboard, mouse, and an infrared receiver for the
> remote control. It does not charge anything. The TV is connected with
> with a DVI to HDMI cable. Audio out goes to an audio receiver.
>
> The PC is an old Shuttle micro ATX box with a Q6600 processor and an AMD
> HD 6350 video board.
>
> I'm not sure I'm ready to load an entirely different OS to test just
> yet. This has been working correctly for a year or 2, and only recently
> stopped starting up correctly from sleep.
>
> It seems that the "sleep" memory is working, since it comes out of sleep
> properly when I push the reset button after it freezes the first time
> after power on. Several seconds after that, the screen is restored
> exactly as it was after being put to sleep using the power button. This
> has got to be the most significant indicator for the problem. Why does
> it not do it correctly when I power it up, but does after a reset.
>
> I tried uninstalling the 3 windows updates installed this month, with no
> change from that.
>
> I tried switching to a different power plan and back, and no change from
> that.
>

Could it be that by using S3 Hybrid, it fails on the first
attempt (RAM issue), then reads the hiberfile on the second
attempt ? It can do that, because it can read the hiberfile
header and discover the file has a valid content. The OS
should be clever enough, to not "pave over" the hiberfile
if it has valid content. But this only happens if you'd
set it up in S3 Hybrid mode, rather than plain S3 mode.

Have you turned off the power to the PC, at the back
for a bit ? Just in case it's a logic gate problem.

That would not clear all the gates.

But, the CMOS well, I don't think it does anything
that would affect the response that late in POST.
I don't think this would need removal of the CMOS
battery. I vaguely recollect there are a few
state bits in the Southbridge, but the power for
that is more likely to be +5VSB related, than
3VSB related.

During sleep, the DIMMs are run in autorefresh mode,
and the power source is +5VSB to the converters
next to the DIMM. My DDR2 machine uses about 5W of
wall power. My DDR3 machine uses 7.5W (but has eight
sticks of RAM), so the power is getting lower with
generation. If I had only two sticks in the DDR3
machine, the +5VSB power would be quite a bit lower.

During POST, the BIOS has to be smart enough to realize
the RAM does not need commissioning, does not need the
special cycle to set CAS value, and so on. The few state
bits in the Southbridge, you read those first to decide
what to do. If the DIMM power converter, next to the DIMMs
were to lose power some time during sleep, then the DIMM
will be uninitialized, and the BIOS POST will not realize
it is insane. Perhaps an attempt to JMP into uninitialized
random DIMM pattern, leads to a double bus fault and a
crash/freeze at BIOS level, and that's when it needs
the reset.

While you can test the RAM with memtest, I don't
think that really helps in this case. Memtest does
have a "storage test" as one of the phases, so it
does have that much working for it. To get closer
to full test coverage, you need to

1) Test two DIMMs at a time.
2) Put the two DIMMs in the same channel,
in single channel mode.
3) Run the complete test cycle at least once.
Shut down. Remove wall plug. Swap high DIMM and
low DIMM on the single channel, in single channel
mode. Power up and run the test again.

By doing that, it provides test coverage of the
BIOS reserved region. The high DIMM has different
reserved locations than the low DIMM, and swapping
the DIMMs changes their role in that configuration.

I think it's just as likely to be a motherboard
related problem, as a bad DIMM problem, because if
the DIMM was bad, you'd probably have had runtime
crashes by now because of it.

You can also try temporarily running the machine
with one DIMM, and see if the symptoms change.

Paul

On 9/28/2021 2:10 PM, Paul wrote:
> Bob F wrote:
>> On 9/28/2021 4:29 AM, Paul wrote:
>>> Bob F wrote:
>>>> I have a Win 8.1 PC that has been working fine for months but now
>>>> will not restart from sleep properly.
>>>>
>>>> When I push the power button to wake it, the power light comes on
>>>> (blue), then it just hangs there and never brings up Win 8.1. The
>>>> disk LED does not flash at all.
>>>>
>>>> If I hold the power button until it turns off and then push it
>>>> again, or just push the reset button, then it starts up just like it
>>>> used to and brings up the previous windows state. The disk LED
>>>> flashes, and the power LED changes to yellow.
>>>>
>>>> It used to restart very quickly from sleep operating from the SSD
>>>> drive.
>>>>
>>>> I have tried:
>>>>     Resetting the BIOS to defaults.
>>>>     Turning off the fast boot function in BIOS - this one causes it
>>>> to display the BIOS splash screen during the hang state.
>>>>     Setting the network card to turn off when the computer is turned
>>>> off.
>>>>     Disabling the "fast restart" in the power options page.
>>>>
>>>> Can anyone offer suggestions to solve this problem?
>>>
>>> You could test with a second OS that supports sleep,
>>> to see if the problem is reproducible (a way to prove
>>> a hardware defect).
>>>
>>> You can also try reducing the number of USB toys
>>> plugged into the PC. Stop charging your iPhone
>>> off the back, as an example of a naughty thing to do.
>>>
>>> The +5VSB that powers the RAM sticks during sleep,
>>> it can't handle more than 2.5A to 3A loading
>>> (check the PSU label for details on +5VSB).
>>>
>>> This power is reserved for PC usage, and is generally
>>> sufficient in most situations.
>>>
>>> However, plugging in a 2 ampere "Apple Load", that
>>> is pushing your luck. The PC could easily be using
>>> 1 ampere to 1.5 amperes for various functions when
>>> the PC is sleeping. An "Apple Load" could push it
>>> over the edge, causing the PSU to temporarily shut
>>> off completely. Then, RAM content is lost.
>>>
>>> While you could try setting up Hybrid Sleep (combines
>>> aspects of S3 and S4, by the S3 also writing out
>>> the Hiberfile just in case), that may not alleviate
>>> the BIOS level problem of "not liking what it sees".
>>>
>>> And that's why I would want the opinion of, say,
>>> a Linux OS where you've gone to the trouble of
>>> getting all the ACPI stuff set up properly. I would
>>> start by Googling for a distro that has out-of-the-box
>>> support for sleep and so on. The last time I looked,
>>> the one I was using wasn't immediately ready to sleep.
>>>
>>>     Paul
>>
>> This PC is being used in my exercise room as a home theater PC. The
>> only USB devices are the keyboard, mouse, and an infrared receiver for
>> the remote control. It does not charge anything. The TV is connected
>> with with a DVI to HDMI cable. Audio out goes to an audio receiver.
>>
>> The PC is an old Shuttle micro ATX box with a Q6600 processor and an
>> AMD HD 6350 video board.
>>
>> I'm not sure I'm ready to load an entirely different OS to test just
>> yet. This has been working correctly for a year or 2, and only
>> recently stopped starting up correctly from sleep.
>>
>> It seems that the "sleep" memory is working, since it comes out of
>> sleep properly when I push the reset button after it freezes the first
>> time after power on. Several seconds after that, the screen is
>> restored exactly as it was after being put to sleep using the power
>> button. This has got to be the most significant indicator for the
>> problem. Why does it not do it correctly when I power it up, but does
>> after a reset.
>>
>> I tried uninstalling the 3 windows updates installed this month, with
>> no change from that.
>>
>> I tried switching to a different power plan and back, and no change
>> from that.
>>
>
> Could it be that by using S3 Hybrid, it fails on the first
> attempt (RAM issue), then reads the hiberfile on the second
> attempt ? It can do that, because it can read the hiberfile
> header and discover the file has a valid content. The OS
> should be clever enough, to not "pave over" the hiberfile
> if it has valid content. But this only happens if you'd
> set it up in S3 Hybrid mode, rather than plain S3 mode.
>
> Have you turned off the power to the PC, at the back
> for a bit ? Just in case it's a logic gate problem.
>
> That would not clear all the gates.
>
> But, the CMOS well, I don't think it does anything
> that would affect the response that late in POST.
> I don't think this would need removal of the CMOS
> battery. I vaguely recollect there are a few
> state bits in the Southbridge, but the power for
> that is more likely to be +5VSB related, than
> 3VSB related.
>
> During sleep, the DIMMs are run in autorefresh mode,
> and the power source is +5VSB to the converters
> next to the DIMM. My DDR2 machine uses about 5W of
> wall power. My DDR3 machine uses 7.5W (but has eight
> sticks of RAM), so the power is getting lower with
> generation. If I had only two sticks in the DDR3
> machine, the +5VSB power would be quite a bit lower.
>
> During POST, the BIOS has to be smart enough to realize
> the RAM does not need commissioning, does not need the
> special cycle to set CAS value, and so on. The few state
> bits in the Southbridge, you read those first to decide
> what to do. If the DIMM power converter, next to the DIMMs
> were to lose power some time during sleep, then the DIMM
> will be uninitialized, and the BIOS POST will not realize
> it is insane. Perhaps an attempt to JMP into uninitialized
> random DIMM pattern, leads to a double bus fault and a
> crash/freeze at BIOS level, and that's when it needs
> the reset.
>
> While you can test the RAM with memtest, I don't
> think that really helps in this case. Memtest does
> have a "storage test" as one of the phases, so it
> does have that much working for it. To get closer
> to full test coverage, you need to
>
> 1) Test two DIMMs at a time.
> 2) Put the two DIMMs in the same channel,
>    in single channel mode.
> 3) Run the complete test cycle at least once.
>    Shut down. Remove wall plug. Swap high DIMM and
>    low DIMM on the single channel, in single channel
>    mode. Power up and run the test again.
>
> By doing that, it provides test coverage of the
> BIOS reserved region. The high DIMM has different
> reserved locations than the low DIMM, and swapping
> the DIMMs changes their role in that configuration.
>
> I think it's just as likely to be a motherboard
> related problem, as a bad DIMM problem, because if
> the DIMM was bad, you'd probably have had runtime
> crashes by now because of it.
>
> You can also try temporarily running the machine
> with one DIMM, and see if the symptoms change.
>
>    Paul

I tried disconnecting the power for a few minutes - no change.

I tried it with 1 dimm, and with the dimms reversed - no change

I re-burned and reset the BIOS - no change

The BIOS battery shows 3.2V

With the BIOS set to show post operations, it is clear that when it does
not start, the BIOS is doing nothing.

If I shut down rather than going to sleep, everything works as expected.

When put to sleep, as it stops, the internal LED on the motherboard goes
off for a moment shortly after the disks and fans stop, then comes back
on. From the sleep mode, the computer seems to usually start up again at
varying times from when it turned off, ending frozen with fan running.

For the moment, I tried going to hibernate instead, and that works fine
with sleep off, so Until I get further inspiration of what to try I will
leave it that way.

Paul - thank you for your help again.

On 9/29/2021 8:53 PM, Bob F wrote:

> I tried disconnecting the power for a few minutes - no change.
>
> I tried it with 1 dimm, and with the dimms reversed - no change
>
> I re-burned and reset the BIOS - no change
>
> The BIOS battery shows 3.2V
>
> With the BIOS set to show post operations, it is clear that when it does not start, the BIOS is doing nothing.
>
> If I shut down rather than going to sleep, everything works as expected.
>
> When put to sleep, as it stops, the internal LED on the motherboard goes off for a moment shortly after the disks and fans stop, then comes back on. From the sleep mode, the computer seems to usually start up again at varying times from when it turned off, ending frozen with fan running.
>
> For the moment, I tried going to hibernate instead, and that works fine with sleep off, so Until I get further inspiration of what to try I will leave it that way.
>
> Paul - thank you for your help again.

In Device Manager (devmgmt.msc), check for keyboard and mouse
Properties having the Power set to "Allow this device to wake the computer".
If you untick all the "wake" items like that, only the power button
on the front will wake from sleep.

If you have a gamer mouse for example, just a slight jiggle of the
table wakes the computer. In theory, if the sleep timer is set to,
say, ten minutes, it would go back to sleep if sleep was working.

I don't know what else to say about your symptoms set.

*******

My typing computer, it's toast today... after it woke from
sleep. All that's left right now, is the empty case is sitting
there, and all the hardware is removed. It all started maybe
three years ago, when seemingly, one of my SATA ports failed,
taking the connected drive with it. A few years passed,
my Marvell NIC died... while asleep. There is something wrong
with sleep power. I put a RealTek NIC in the machine, and
all was fine. I was limping along with five SATA ports and
a cheap NIC.

Now, I made a mental note "do not sleep this machine, ever again",
as the symptom set now points at sleep power as a problem. In
a moment of dumbness last night, I put the machine to sleep.
Today, all it will do for me is freeze, so... "off with its head".

I can fix the odd thing here, but when your number is up,
that's it. That was the third strike. The motherboard is a mass
of heatsinks, some of which are riveted to the motherboard.
This means I can't visually inspect the whole thing.

Paul

On 9/29/2021 6:36 PM, Paul wrote:
> On 9/29/2021 8:53 PM, Bob F wrote:
>
>> I tried disconnecting the power for a few minutes - no change.
>>
>> I tried it with 1 dimm, and with the dimms reversed - no change
>>
>> I re-burned and reset the BIOS - no change
>>
>> The BIOS battery shows 3.2V
>>
>> With the BIOS set to show post operations, it is clear that when it
>> does not start, the BIOS is doing nothing.
>>
>> If I shut down rather than going to sleep, everything works as expected.
>>
>> When put to sleep, as it stops, the internal LED on the motherboard
>> goes off for a moment shortly after the disks and fans stop, then
>> comes back on. From the sleep mode, the computer seems to usually
>> start up again at varying times from when it turned off, ending frozen
>> with fan running.
>>
>> For the moment, I tried going to hibernate instead, and that works
>> fine with sleep off, so Until I get further inspiration of what to try
>> I will leave it that way.
>>
>> Paul - thank you for your help again.
>
> In Device Manager (devmgmt.msc), check for keyboard and mouse
> Properties having the Power set to "Allow this device to wake the
> computer".
> If you untick all the "wake" items like that, only the power button
> on the front will wake from sleep.

That is actually what I want and how I have set it.

>
> If you have a gamer mouse for example, just a slight jiggle of the
> table wakes the computer. In theory, if the sleep timer is set to,
> say, ten minutes, it would go back to sleep if sleep was working.
>
> I don't know what else to say about your symptoms set.
>
> *******
>
> My typing computer, it's toast today... after it woke from
> sleep. All that's left right now, is the empty case is sitting
> there, and all the hardware is removed. It all started maybe
> three years ago, when seemingly, one of my SATA ports failed,
> taking the connected drive with it. A few years passed,
> my Marvell NIC died... while asleep. There is something wrong
> with sleep power. I put a RealTek NIC in the machine, and
> all was fine. I was limping along with five SATA ports and
> a cheap NIC.
>
> Now, I made a mental note "do not sleep this machine, ever again",
> as the symptom set now points at sleep power as a problem. In
> a moment of dumbness last night, I put the machine to sleep.
> Today, all it will do for me is freeze, so... "off with its head".
>
> I can fix the odd thing here, but when your number is up,
> that's it. That was the third strike. The motherboard is a mass
> of heatsinks, some of which are riveted to the motherboard.
> This means I can't visually inspect the whole thing.
>
> Paul
>

Thanks for the motivation to maybe let go of the sleep problem.

Subject	Author
New problem with starting PC from sleep mode	Bob F
Re: New problem with starting PC from sleep mode	Paul
Re: New problem with starting PC from sleep mode	Bob F
Re: New problem with starting PC from sleep mode	Paul
Re: New problem with starting PC from sleep mode	Bob F
Re: New problem with starting PC from sleep mode	Paul
Re: New problem with starting PC from sleep mode	Bob F
Re: New problem with starting PC from sleep mode	Mr. Man-wai Chang

"Any excuse will serve a tyrant." -- Aesop

computers / alt.comp.hardware.pc-homebuilt / New problem with starting PC from sleep mode

computers / alt.comp.hardware.pc-homebuilt / New problem with starting PC from sleep mode