All these years I've been avoiding Win10. Now I'm finally
setting up a new system. I was in the middle of copying
over a Raspberry Pi disk image (a 1 hour project) when I
got a BSOD and it rebooted. Boy was I surpised! I don't
think I've seen a BSOD since Win98. No one told me that
MS brought them back. If I'd known that MS had brought
back BSOD I might have switched to Win10 earlier. And they
force you to reboot. How cool is that? :)

It's so nostalgic, after years on XP and 7, to finally have a
Windows system that might fall apart any any time. It
brings back memories of installing Mindspring from floppies.

Newyana2 <Newyana2@invalid.nospam> wrote:

> All these years I've been avoiding Win10. Now I'm finally
> setting up a new system. I was in the middle of copying
> over a Raspberry Pi disk image (a 1 hour project) when I
> got a BSOD and it rebooted. Boy was I surpised! I don't
> think I've seen a BSOD since Win98. No one told me that
> MS brought them back. If I'd known that MS had brought
> back BSOD I might have switched to Win10 earlier. And they
> force you to reboot. How cool is that? :)
>
> It's so nostalgic, after years on XP and 7, to finally have a
> Windows system that might fall apart any any time. It
> brings back memories of installing Mindspring from floppies.

"copying over a Raspberry Pi disk image"

There's your clue. Drivers run a kernel level, and must be designed for
the OS under which they are installed, and must match the bitwidth of
the OS under which they run.

"Raspberry Pi" is hardware. Can run multiple operating systems. You
didn't mention which image on which model of Pi you copied over to which
hardware running Win10. I would think Pi hardware would look alien to
Intel or AMD hardware for your Intel or AMD Win10 computer. Does
Raspberry guarantee their image of Raspberry Pi OS (aka Raspbian) is
100% compatible with Windows 10 installs, or even on non-Pi hardware?

https://www.raspberrypi.com/software/
"Debian with Raspberry Pi Desktop is our operating system for PC and
Mac. It provides the Raspberry Pi OS desktop, as well as most of the
recommended software that comes with Raspberry Pi OS, for any PC or
Apple Mac computer."

Doesn't look like Raspbian runs as a subordinate OS under Windows 10.
If you want it and Win10 on the same desktop, perhaps you have to use
virtual machines to keep them separate. Maybe you got some version of
Windows to run on the Pi hardware, but the drivers in that OS won't
match the drivers for Windows on an Intel/AMD desktop.

https://www.easeus.com/partition-manager-software/install-windows-10-on-raspberry-pi-4.html

That describes how to use Win10 on Pi hardware. That's not the same as
taking an Win10 image on Pi hardware thinking the image will work for
Win10 on whatever hardware is in your "new system".

BSODs have error codes. You didn't mention yours. Have you yet visited
the Event Viewer to check for critical errors?

VanguardLH wrote:

> "copying over a Raspberry Pi disk image"
>
> There's your clue.
Writing a rPi image using a Windows machine involves taking a single
large file (e.g. 1.1GB for the latest raspiOS) and writing it sector by
sector to an SD card

Now, what makes you think that Windows should fail to properly write it
based on what the file contains? windows doesn't care what's in the
sectors, or what the bitwidth is, its just data.

Who needs the clue?

On 2/12/2024 3:44 AM, Andy Burns wrote:
> VanguardLH wrote:
>
>> "copying over a Raspberry Pi disk image"
>>
>> There's your clue.
> Writing a rPi image using a Windows machine involves taking a single large file (e.g. 1.1GB for the latest raspiOS) and writing it sector by sector to an SD card
>
> Now, what makes you think that Windows should fail to properly write it based on what the file contains?  windows doesn't care what's in the sectors, or what the bitwidth is, its just data.
>
> Who needs the clue?
>

Normally, I would suggest a memtest floppy as a starting point.
But the bottom of this post, shows a candidate source of grief.
Likely no need to test RAM now. I would be using my portable
USB to SD stick right about now.

*******

Well, how could a tool like this crash the system ? It's Windows! :-) That's how.
The platform used is QT, for its cross-platform goodness (the ability to deploy
on three platforms). I will pretend it uses QT5, without actual proof of that.

https://www.raspberrypi.com/news/raspberry-pi-imager-imaging-utility/

https://www.raspberrypi.com/software/

Install Raspberry Pi OS using Raspberry Pi Imager

Download for Windows <=== imager_1.8.5.exe
Download for macOS <=== .dmg (DiskImage mounter)
Download for Ubuntu for x86 <=== .deb

Name: imager_1.8.5.exe
Size: 20,268,248 bytes (19 MiB)
SHA256: 659C54979FA4C75840E4EE9B17393BE08DD86C5E7C726493B7EB58A8623BB6A7

Not much info of note. Program looks like it could be QT based.
It installs (Nullsoft installer is mentioned).
https://www.virustotal.com/gui/file/659c54979fa4c75840e4ee9b17393be08dd86c5e7c726493b7eb58a8623bb6a7/details

I could find an example of bad behavior here, with a QT application appearing
to blow something up. You'd think Windows would harvest the puppy.

https://forum.qt.io/topic/109163/qt-deployed-executable-crashes-on-some-computers/42

*******

Encourages by finding QT is fallible, I try a focused search, and there is a candidate.

https://github.com/raspberrypi/rpi-imager/issues/141

"Previously people have reported similar problems with Dell laptops that come
with an internal Realtek SD card reader that uses custom drivers. <==== OK, Ring 0 country, check!
(Most other SD card readers do not use custom drivers, but use the <==== A dynamite factory.
generic USB mass storage drivers written by Microsoft instead)
See: #94
"

Which means using a USB stick with the SD hole in the side, might
work better... as long as it's not the same RealTek chip in a party dress.

Where would we be without RealTek.

Paul

Newyana2,

> finally have a Windows system that might fall apart any time.

You can ofcourse point your finger at the OS, but I would suggest to
check/test your hardware.

A BSOD /can/ be caused by malfunctioning software*, but is often caused by
aging hardware (which develops faults over time). Heck, even simple
over-heating can cause it.

* what (external?) driver are you using to access the RPi (Linux-style
formatted) drive ? Such drivers are low enough in a system to be able to
cause BSODs when they access memory they do not own.

Example: On XP I had FF crash and wondered if one of the add-on's I wrote
was wonky. It turned out that the memory controller was failing (the memory
itself was fine).

Also, if a program (driver?) has got a memory leak you often won't even know
about it, untill you try to do something big (an hour of copying?) and the
accumulated leaks could result in the system running outof resources,
causing a (not to-well error-checked) program to do something "it should not
have done".

> And they force you to reboot.

:-) What other option should it have presented ? How do you think the OS
could recover from a BSOD ? Its not called "Of Death" for nothing you know.

Also, the "reboot after BSOD" is the default response - to give a 9-to-5
user back a working 'puter. However, you can switch the automatic reboot
off, allowing you to actually read the info on the BSOD screen (and not have
to dig into the log files).

Regards,
Rudy Wieser

"R.Wieser" <address@is.invalid> wrote

| You can ofcourse point your finger at the OS, but I would suggest to
| check/test your hardware.
| | A BSOD /can/ be caused by malfunctioning software*, but is often caused by
| aging hardware (which develops faults over time). Heck, even simple
| over-heating can cause it.
|

The computer is a week old. :)
But in fairness to Win10 (though I can't forgive MS for making
blue screens time out before I can read them), I decided to do some
checks. SFC reported fine. Memtest86 reported errors. Microsoft's
useless tester reported only "hardware" errors and said I should "contact
the computer maker".

I took out a RAM stick and then it tested fine. I carefully vacuumed
out the RAM slot, put the RAM back, and now it's all testing fine. So...
Maybe the RAM was a bit loose? But I've been setting up the machine
for 2 weeks without problems.

Now I'm wondering if there are other possible issues. Loose connections?
Everything seems to point to a bad hardware connection with the second
stick, but I don't see how that took so long to pop up.

| * what (external?) driver are you using to access the RPi (Linux-style
| formatted) drive ? Such drivers are low enough in a system to be able to
| cause BSODs when they access memory they do not own.
| The RPi has nothing to do with it. I was just reading the data from an
SD card in a USB stick reader, to make an ISO of the RPi system. I was
using a program called Win32 Disk Imager.

"Paul" <nospam@needed.invalid> wrote

|
| Normally, I would suggest a memtest floppy as a starting point.
| But the bottom of this post, shows a candidate source of grief.
| Likely no need to test RAM now. I would be using my portable
| USB to SD stick right about now.
|

I think the RPi issue is a red herring. I was using a Windows
program designed to work with a USB SD adapter. After dealing
with the RAM issue I had no trouble reading the Pi SD and writing
the disk image to two other SDs for backup.

As I explained to Rudy, Memtest86 told me I had mucho RAM
errors. (As in "F0000000 Expected F0000020")
I took out the second stick and retested. Fine. I carefully
put back the second stick. Fine. So far, so good. So that sounds
like a loos stick, right? But I'm still a bit nervous, as no problems
arose during extensive install and config operations in setting up
the system. I've been working on it quite a bit for 2 weeks now,
after having built it. (MSI B760M-P Pro motherboard. G.Skill Ripjaws
dual channel pair of 8 GB DDR4-3200 RAM.)

Any thoughts on that scenario?

For many years there has been information on how to stop auto start
after a BSOD

KenW

"KenW" <ken1943@invalid.net> wrote

| For many years there has been information on how to stop auto start
| after a BSOD
| Ah. Since you didn't bother to explain, I researched
it so that others can benefit. It turns out that these options
have existed at least since XP, but MS apparently changed the
defaults somewhere along the line. I was never aware of the
options.

Settings->System->About->Advanced System Settings.
Click Settings under "Start Up and Recovery" on the Advanced tab.
Uncheck Automatic Restart.

I haven't checked my Win10 system yet, but looking at it here
on XP I see there's also an option to write a log, which is enabled
by default on XP, while the auto restart is not. I also have it
configured by default to write a debugging log. Nice to know.

Newyana2 wrote:
> The computer is a week old. 🙂
So not yet proven reliable ...

Newyana2,

> The computer is a week old. :)

Arghh... Yeah, that pretty-much rules out failures due to age. :-)

> I took out a RAM stick and then it tested fine. I carefully
> vacuumed out the RAM slot, put the RAM back, and now it's all
> testing fine. So... Maybe the RAM was a bit loose?

Or just a tad bit corroded (connector or RAM), and re-seating it caused the
connector to cut thru that sheen of corrosion enough to make a good contact.

> Everything seems to point to a bad hardware connection with the
> second stick, but I don't see how that took so long to pop up.

I mentioned my FF crashing. My 'puter seemed to work allright most of the
time. Possibly (likely) because it just didn't reach the wonky address
space.

I did notice a few other (written by me) programs crashing, but just as with
FF I suspected my coding (and almost pulled my hair out when I could not
find any mistakes).

> The RPi has nothing to do with it. I was just reading the data from
> an SD card in a USB stick reader, to make an ISO of the RPi system.
> I was using a program called Win32 Disk Imager.

Ah, yes. I was imagining file-based backup-ing (which I do myself), and
forgot all about that possibility. :-|

I have to ask though : how come you suspected the OS, which (I suspect) you
had been using without a problem for a number of years, and not the "big
change" you just made, the computer ?

Regards,
Rudy Wieser

Newyana2 <Newyana2@invalid.nospam> wrote:
> "R.Wieser" <address@is.invalid> wrote
>
> | You can ofcourse point your finger at the OS, but I would suggest to
> | check/test your hardware.
> |
> | A BSOD /can/ be caused by malfunctioning software*, but is often caused by
> | aging hardware (which develops faults over time). Heck, even simple
> | over-heating can cause it.
> |
>
> The computer is a week old. :)
> But in fairness to Win10 (though I can't forgive MS for making
> blue screens time out before I can read them),

Well, that's what your smartphone is for, you just take a picture! :-)

But seriously, as Rudy mentioned, you can configure it to *not* reboot
automatically after a BSOD:

'System Properties' applet (C:\Windows\System32\sysdm.cpl) ->
'Advanced' tab -> Start-up and Recovery -> Settings... -> System failure
-> untick 'Automatically restart' -> OK -> OK.

[...]

"R.Wieser" <address@is.invalid> wrote

| > I took out a RAM stick and then it tested fine. I carefully
| > vacuumed out the RAM slot, put the RAM back, and now it's all
| > testing fine. So... Maybe the RAM was a bit loose?
| | Or just a tad bit corroded (connector or RAM), and re-seating it caused
the
| connector to cut thru that sheen of corrosion enough to make a good
contact.
|

That's what I'm hoping. I hope it's not a loose solder or some
such on the motherboard. But at this point there's no reason
to suspect bigger problems. Yet it just seems odd that it took 2
weeks to see a crash. RAM problems make me nervous. Things
can get very corrupt by the time the problem is worked out.

|
| I have to ask though : how come you suspected the OS, which (I suspect)
you
| had been using without a problem for a number of years, and not the "big
| change" you just made, the computer ?
| That was meant half as a joke. I was just making a dig at
Win10, because I can't remember the last time I saw a BSOD
on XP. It's been years.

But I'm also not very experienced with Win10, and so far it seems
surprisingly brittle to me. So I wasn't sure that the problem wasn't just
Win10 itself. That's the non-joke half of my complaint. I'm just getting
settled in and after 2 weeks I still don't feel ready to make the move.
Win10 seems to be still burping up noxious gases. And I've installed
Visual Studio 6 without problems, but I don't yet know whether that
will have problems.

I do have a Win10 laptop that I use for a few things, like websites
that won't work on XP. But I haven't tried to make that my primary
computer, so I'd been putting up with nags and junk and a vast
number of nonsense entries in the Explorer folder tree. And I haven't
tried to use it for much beyond browsing.

An example of what seems like brittleness: I now have a behavior
where the Personalization applet opens to the blue box with the white gear
and hangs there, for maybe a minute or two. Running Procmon I don't
see anything suspicious going on during that time. And this is on a
brand new system. It started happening with no action I'm aware of
on my part. I did put in a new Desktop picture, but that was before the
hang started happening. It's not a big problem, but it is mysterious and
gives me pause about depending on Win10.

Interestingly, every problem I've had turns out to be common, with
lots of discussion online. I guess there's an advantage to coming to an
OS 10 years late. All the tweaks and bug fixes are worked out, so I
don't have to write them myself. :) (Although while lots of people have
had the same Personalization hang, I didn't find a fix.)

Win10 just seems to be generally less stable and less cooperative
than earlier systems. The UI choices, for example... I can no longer
choose all the window colors. What used to be the menu or button color,
also known as 3D objects color, is putty in Firefox, almost white
in Explorer, gray in my own software... I haven't figured out where
that's coming from, and Win10 won't let me pick it. The taskbar,
which used to be button color, is now matched to the title bar color.
WinAero Tweaker gave me access to the inactive titlebar color. But
so far that seems to be all that I can choose. I would have liked to
have had light gray window frames. There's too much white in the
windows, making it slightly confusing to distinguish the frames from
the active areas.

Newyana2,

> | I have to ask though : how come you suspected the OS, which (I suspect)
> | you had been using without a problem for a number of years, and not the
> | "big change" you just made, the computer ?
>
> That was meant half as a joke.

Ah. :-)

> I was just making a dig at Win10, because I can't remember the last time
> I saw a BSOD on XP. It's been years.

I had one "recently" (a year or two back) on the 'puter with the memory
controller problem. After having replaced the motherboard (and nothing
else) I haven't had another problem. A BSOD before that ? It must have
been a so far back that I can't even remember anymore.

> But I'm also not very experienced with Win10,

Same here - by choice.

> Interestingly, every problem I've had turns out to be common,
> with lots of discussion online. I guess there's an advantage
> to coming to an OS 10 years late. All the tweaks and bug fixes
> are worked out, so I don't have to write them myself. :)

:-) Thats the "pre" of not wanting/needing to be part of the cutting edge.

> Win10 just seems to be generally less stable and less cooperative
> than earlier systems.

I don't know about it being unstable (pretty-much zero experience), but have
heard a number of people complain about that either not being able to find
configuration settings, or that they simple do not exist (anymore).

Personally I would not have much of a problem if they would keep updating
the OSes innards, as long as they would not forever be changing the UI and
(the location of) the settings.

Athough I well understand the companies need to create new "bling" - money
needs to keep flowing in, otherwise its will die - I have zero wish to run
that rat-race.

Regards,
Rudy Wieser

Andy Burns <usenet@andyburns.uk> wrote:

> VanguardLH wrote:
>
>> "copying over a Raspberry Pi disk image"
>>
>> There's your clue.
>
> Writing a rPi image using a Windows machine involves taking a single
> large file (e.g. 1.1GB for the latest raspiOS) and writing it sector by
> sector to an SD card
>
> Now, what makes you think that Windows should fail to properly write it
> based on what the file contains? windows doesn't care what's in the
> sectors, or what the bitwidth is, its just data.
>
> Who needs the clue?

He got a new computer where he is running Windows 10 (to be on-topic
here). Raspberry hardware is discussed over in comp.sys.raspberry-pi.
That's not the hardware discussed here for Win10 systems. How is the
hardware for his "new system" the same as for his Pi hardware? He never
mentioned anything about his "new system", but I've not seen anyone
discussing Pi here before.

Is "copying over a Raspberry Pi disk image" not the same as creating an
image backup of the OS partition, and then expecting foreign hardware to
run that image on different hardware?

The Raspberry Pi Imager utility runs on Windows, MacOS, or Ubuntu,
because an OS is needed to run the utility to download and install
Raspbian in the SD card on the Pi board. The OP didn't say how he was
"copying" a Pi image (presumably from the Pi board to his Windows 10
"new system") nor did he mention the Pi Manager tool. You see a lot of
details in his post? THAT is all the info initial responders have to
work with.

Newyana2 <Newyana2@invalid.nospam> wrote:

> But in fairness to Win10 (though I can't forgive MS for making
> blue screens time out before I can read them),

Configure Windows to /not/ reboot on a BSOD. It's stupid to have the OS
keep rebooting, especially since it could crash again, and the computer
is stuck in continual reboot cycling. Run sysdm.cpl, Advanced tab,
Startup and Recovery settings button. Under System Failure section,
disable the "Automatically restart" option. You might also want to
change to a small dump for debug info unless you are an OS programmer
delving into a crash that was logged (assuming you have the "Write an
event to the system log" enabled).

> I took out a RAM stick and then it tested fine. I carefully vacuumed
> out the RAM slot, put the RAM back, and now it's all testing fine.
> So... Maybe the RAM was a bit loose? But I've been setting up the
> machine for 2 weeks without problems.

A problem with repeatedly powering up to use a computer and powering
down when done is thermal creep. Thermal expansion and contaction can
make components is non-secured connectors (just friction holds them
together) creep apart. I dust out my computer once per year. That's
when I remove and reinsert the memory modules, and any other connections
that are held together merely by friction. Disconnecting and
reconnecting also wipe clean the contacts.

> Now I'm wondering if there are other possible issues. Loose connections?
> Everything seems to point to a bad hardware connection with the second
> stick, but I don't see how that took so long to pop up.

I've seen memory that tested okay with memtest86, but failed under
stress. memtest86's basic "is it okay" diagnostic testing isn't a
stress test, just a validation test. Did you run its burn-in, SIMD, and
hammer tests? There is the default simple validation test, and then
there are its burn-in,

Notice what they say about incomplete cell testing due to chip density:

https://www.memtest86.com/tech_memtest-algoritm.html

Did the [default] test you ran of memtest86 include all the test
algorithms mentioned at:

https://www.memtest86.com/tech_individual-test-descr.html

I don't know if the default test includes all those algorithms. Also,
from their descriptions, they really are not stress tests. I think
Prime95 is still a good stress and stability tests on CPU and memory.
Been awhile, but I recall where memtest86 passed all memory, but Prime95
failed. There's a difference in raw testing the memory versus stress
testing memory via the OS. Prime95 and memtest64 (that's 64, not 86)
run under the OS. memtest86 is booted, ran standalone, so it doesn't
run under your OS.

Are you overclocking the memory? There's its true rating, and there's
the clocking it might endure when in the mobo at whatever is the
clocking configured there. Some systems/mobos even come with
overclocking as the default. If the new system is a pre-built, you
don't really know what is the true clocking, CAS, and other specs to see
if they match up with the BIOS settings for memory config. You could
use CPU-Z to read the memory's SPD settings.

https://www.memtest86.com/tech_ram-spd.html

I have not use memtest86 in many years, so I don't recall if it uses the
BIOS memory settings or reads the SPD and runs at the true specs. If it
did the latter, it may not be testing at the same settings as when the
memory is used under the OS and apps.

> The RPi has nothing to do with it. I was just reading the data from an
> SD card in a USB stick reader, to make an ISO of the RPi system. I
> was using a program called Win32 Disk Imager.

Not the Raspberry Pi Imager tool? That lays an image on the SD card for
the Raspbian OS.

https://projects.raspberrypi.org/en/projects/raspberry-pi-getting-started/1

Maybe that's not what you want, and instead want to create an .iso image
of what is already on the Pi's SD card. My guess what is the "Win32
Disk Imager" is the one at:

https://sourceforge.net/projects/win32diskimager/

Was it when using that imaging program that you got the BSOD in Windows
10? If so, I'd look at the Event logs to see what was getting reported
as the source of the problem. If you configure Windows to save a
[small] mini-dump file on crashes (as long as the crash isn't so low
level that events don't get recorded by the OS), you can use Nirsoft's
BlueScreeView (https://www.nirsoft.net/utils/blue_screen_view.html) to
give a bit more comprehensible description of the event. Just be aware
that the top-level error may not be the cause, so you have to drill down
to the cause that was exposed higher up.

Be aware that even if you configure Windows to save minidumps, there are
cleanup tools that will delete the dump files. For example, CCleaner
under its Custom Clean -> System category can delete dump files, if you
enable it. Minidumps are only 64KB in size, so you could keep many.

https://learn.microsoft.com/en-us/windows-hardware/drivers/debugger/small-memory-dump

Not sure what you mean by a cable to a USB card reader. You ran a USB
cable from a USB port in your new Win10 system to an external SD card
reader? Is the USB stick reader only to support the card reader, or
does it contain a USB hub, too? I've run into problems using USB hubs,
even powered ones to avoid overloading the hub with too many concurrent
connections (and get a powered USB hub that can support the full load on
EVERY one of its ports, not just some of them at the same time). Even
then I've had issues with hubs, like when used between a media playback
device to the hub to the USB HDD where the movie files are stored. Had
to get rid of the hub, and plug the USB HDD directly into the media
player's USB port. I really didn't need the hub. Had an expectation
that I might need multiple USB HDDs to store all my movies, but instead
got a huge USB HDD. The hub had switches to enable/disable its output
ports to prevent power overloading from just the media player's USB
port. Haven't played with other USB hubs to see if the faults were in
the one I choose, but then just the one USB HDD directly to media
player's USB port works, and only half of it is used, so far.

"R.Wieser" <address@is.invalid> wrote

| I don't know about it being unstable (pretty-much zero experience), but
have
| heard a number of people complain about that either not being able to find
| configuration settings, or that they simple do not exist (anymore).
| That's true. Things like Display color options have
disappeared. (HKEY_CURRENT_USER\Control Panel\Colors
settings still exist, but seem to have no effect. Apparently
the staff at MS just overwrote all that without removing
the code.)

And the settings can be maddening. There are two different
sets. One is the Control Panel. The other is a PC Settings
window that looks Metro-esque and contains a subset of
what's in CP, along with other things that are not in CP. So
one can go in circles trying to find things. I've also found that
much of the online info doesn't line up. For example, sites online
told me that I could choose to disable LUA (the portion of UAC
that still operates when UAC settings are at their lowest levels).
But there's no such option in the window and I have the most
recent version of Win10.

I got into the LUA thing because I noticed that one of my
programs had broken file drag/drop on Win10, but not on another
Win10 box. Huh? It turned out to be this secret LUA setting.
With LUA enabled, drag-drop to an elevated window won't work,
even though it's me dropping the file to the program that me
is running. Nutty stuff.

Newyana2 <Newyana2@invalid.nospam> wrote:

> But I'm also not very experienced with Win10, and so far it seems
> surprisingly brittle to me.

Did you let Windows Update (WU) install the hardware drivers, or did you
get the drivers from the hardware manufacturers own web site? WU
doesn't always pick the best driver (and that is an old problem, not
just with Win10), and often biases to new versions of a driver instead
of stable versions. I disable hardware driver updating in Windows 10,
and *I* decide if and when to get them from the hardware makers' web
sites. Most drivers come with the mobo, or you visit the mobo maker's
web site to get them when setting up a new system. Even when drivers
come with the mobo (e.g., on CD media), and after getting the system
working, I visit the mobo maker's web site to see if there is a later
version of their driver package. I even check if there is a later
version of BIOS firmware, but I don't update unless I hit a problem, or
there was an added feature that I want.

For example, WU will try to update the Intel network drivers on my
setup. No, WU is disabled from doing hardware updates. I could use
Intel's own update checker, but I just do it myself. Some folks don't
like to do research on drivers to see if a new version applies to their
particular hardware, or if it even addresses issues that are exhibited
in your setup.

Another example is when a hardware component has multiple versions, but
the hardware changes between hardware versions. WU will pick the latest
hardware driver for a "family" of components, but can pick a driver for
a later hardware config than the older model you have. Back when I had
a Winmodem, the card maker changed the DSP on later versions of their
hardware, and WU picked a hardware driver for the latest hardware
version, but my Winmodem has the old DSP, so the new driver was not the
correct choice.

I could point at various online articles on how to disable hardware
updating by WU. Instead I just use WinAero Tweaker to disable hardware
updates. In WinAero, and instead of hunting through all its settings,
search on "update", and select "driver updates". It also makes it easy
to disable WU until if and when you are prepared to change the state of
your OS installation, like when you have the time, initiative, have
saved an image backup of the OS partition(s), do the updates, test for
problems, and save an image backup of the changed partition(s).

Note: When you disable WU, some features become unavailable. For
example, you cannot view the history of updates. WU is disabled, so
Windows won't let you see the update history. You can still see the
update history using Nirsoft's Windows Updates View.

> And I've installed Visual Studio 6 without problems, but I don't yet
> know whether that will have problems.

Just curious: Was that the free Community edition, or the paid edition?
VS6 sounds old (released in 1998 and last version to support Win9x), but
I don't know what version you get with the Community edition (maybe
2022). I would think VS6 wouldn't support latter .NET Framework
versions, later C runtime libs, and features in later versions of
Windows (2000, XP, Vista, 7, 9, 10, 11). VS6 was the last version that
still had Visual J++ which they had to remove in later VS versions in a
settlement with Sun Microsystems. VS6 was released back in 1998. Lots
of changes since then in Windows post-9x/ME.

> An example of what seems like brittleness: I now have a behavior where
> the Personalization applet opens to the blue box with the white gear
> and hangs there, for maybe a minute or two. Running Procmon I don't
> see anything suspicious going on during that time. And this is on a
> brand new system. It started happening with no action I'm aware of on
> my part. I did put in a new Desktop picture, but that was before the
> hang started happening. It's not a big problem, but it is mysterious
> and gives me pause about depending on Win10.

Same happens when starting Win10 in its Safe Mode?

Whose video hardware are you using: the integral Intel CPU video, or a
video card in a daughercard slot? Did you match the video driver to the
hardware source? Was the "new system" a pre-built, or did you build it?

"VanguardLH" <V@nguard.LH> wrote

| Configure Windows to /not/ reboot on a BSOD.

Yes. See KenW's post above. I didn't know it was an optional
setting, but now I do and I disabled the rebooting. I'm also going
to disable the memory dump. Windows wrote a 700MB file!
It appears to be just raw memory data -- nothing that I have
the expertise to make sense of.

| > I took out a RAM stick and then it tested fine. I carefully vacuumed
| > out the RAM slot, put the RAM back, and now it's all testing fine.
| > So... Maybe the RAM was a bit loose? But I've been setting up the
| > machine for 2 weeks without problems.
| | A problem with repeatedly powering up to use a computer and powering
| down when done is thermal creep. Thermal expansion and contaction can
| make components is non-secured connectors (just friction holds them
| together) creep apart. I dust out my computer once per year. That's
| when I remove and reinsert the memory modules, and any other connections
| that are held together merely by friction. Disconnecting and
| reconnecting also wipe clean the contacts.
| The thing is, though, that I just built the computer 2 weeks ago.
It's hard to click in RAM sticks without proper contact, but it's possible
that was the problem.

| I've seen memory that tested okay with memtest86, but failed under
| stress. memtest86's basic "is it okay" diagnostic testing isn't a
| stress test, just a validation test. Did you run its burn-in, SIMD, and
| hammer tests? There is the default simple validation test, and then
| there are its burn-in,
|

The hammer test was part of it. But frankly I just let Memtest do
its thing. In any case, when I first tested, with both sticks in, there
were hundreds of errors recorded. With one stick out it passed. When
I put the stick back it passed. So none of that was under unusual load.

One unusual thing with Memtest: It said there are 12 cores, with
6 active. Is it possible that I actually bought a much higher powered
CPU and they've crippled it to sell it down-market?

| Notice what they say about incomplete cell testing due to chip density:
| | https://www.memtest86.com/tech_memtest-algoritm.html
| | Did the [default] test you ran of memtest86 include all the test
| algorithms mentioned at:
| | https://www.memtest86.com/tech_individual-test-descr.html
| | I don't know if the default test includes all those algorithms. Also,
| from their descriptions, they really are not stress tests. I think
| Prime95 is still a good stress and stability tests on CPU and memory.

Interesting. I hadn't heard of that one.

| Are you overclocking the memory? There's its true rating, and there's
| the clocking it might endure when in the mobo at whatever is the
| clocking configured there.

I'm not doing any fancy business. I use a computer in a practical
way and have never played games to speak of. I'd have no use for
overclocking anything. As far as I know, my motherboard has no options
for overclocking. It's just basic. I bought something a few years old
to save money. A 12th gen i5 CPU and hardware to go with it. For me
there's no sense going for speed. It's already way faster than what I need.

| > The RPi has nothing to do with it. I was just reading the data from an
| > SD card in a USB stick reader, to make an ISO of the RPi system. I
| > was using a program called Win32 Disk Imager.
| | Not the Raspberry Pi Imager tool? That lays an image on the SD card for
| the Raspbian OS.
| I tried that but it seemed unrespoonsive, and other sites
other than the Raspbian people recommended the Win32 Disk Imager,
which worked well. I copied an image to disk, wrote that back to a
new SD card, put that into the Pi, and it works fine.

| Was it when using that imaging program that you got the BSOD in Windows
| 10? If so, I'd look at the Event logs to see what was getting reported
| as the source of the problem.

Nothing in the event log. At the time it crashed, the
copy of the disk image was about halfway done, I was running
FF, and I was taking a file out of a ZIP to read. So, not a lot
of strain.

| If you configure Windows to save a
| [small] mini-dump file on crashes (as long as the crash isn't so low
| level that events don't get recorded by the OS), you can use Nirsoft's
| BlueScreeView (https://www.nirsoft.net/utils/blue_screen_view.html) to
| give a bit more comprehensible description of the event. Just be aware
| that the top-level error may not be the cause, so you have to drill down
| to the cause that was exposed higher up.

Thanks. I'll look into that. Nirsoft is such a remarkably useful
resource. All of that softaware that I've used has been lightweight,
efficient and very sensible.

| Not sure what you mean by a cable to a USB card reader. You ran a USB
| cable from a USB port in your new Win10 system to an external SD card
| reader?

I didn't say cable. It's a nifty little gizmo that came with the Pi --
A USB stick with a tiny slot for the SD card. So I working with the
SD card but Windows sees it as a USB stick. There are no complications
there. No USB hub. Just a slot in the computer. I know what you
mean with hubs. If they don't have their own power then some
things won't work in them.

Newyana2 wrote:

> I'm also going to disable the memory dump. Windows wrote a 700MB
> file! It appears to be just raw memory data -- nothing that I have
> the expertise to make sense of.

WinDbg (and other tools) can analyse .dmp files, useful if it's e.g. a
reproducible driver issue, not so much if it's caused by random memory
faults.

"VanguardLH" <V@nguard.LH> wrote

| Did you let Windows Update (WU) install the hardware drivers, or did you
| get the drivers from the hardware manufacturers own web site?

V! What do you take me for? :) I always go to the manufacturer.
This is the first time the MB didn't come with a disk, but I found
the drivers online without trouble.

| I could point at various online articles on how to disable hardware
| updating by WU. Instead I just use WinAero Tweaker to disable hardware
| updates.

That's a very nice program.

| > And I've installed Visual Studio 6 without problems, but I don't yet
| > know whether that will have problems.
| | Just curious: Was that the free Community edition, or the paid edition?

Paid. Though back when I bought it there were reasonably good deals.

| VS6 sounds old (released in 1998 and last version to support Win9x), but
| I don't know what version you get with the Community edition (maybe
| 2022). I would think VS6 wouldn't support latter .NET Framework
| versions, later C runtime libs, and features in later versions of
| Windows (2000, XP, Vista, 7, 9, 10, 11). VS6 was the last version that
| still had Visual J++ which they had to remove in later VS versions in a
| settlement with Sun Microsystems. VS6 was released back in 1998. Lots
| of changes since then in Windows post-9x/ME.
| VS6 is pre-dotnet, yes. I work with VB6, which still writes software
that's supported on virtually every Windows box in existence. Non .Net
bloat. No .Net limited compatibility. None of that crap. VB6 is nice because
it has the convenience of RAD -- like drag-drop window components --
but it can also handle Win32 API and call into most DLLs. So I use
the GUI simplicity and mostly API code.

VB6 lacks some of the newer conveniences. For example, awhile back
I wanted to be able to download webbpages via https. I'd originally
written raw winsock code. Later I switched to winhttp. But neither of
those could handle encryption, which gets very complicated, beyond my
skill level. I eventually found libcurl, which works fine. I'm sure that
dotnet probably wraps https in one line:
GoGetMe "https://somewhere.com/index.html

But that comes at a cost. And figuring these things out is fun. I suppose
I'd talk differently if I were being paid by the hour. Then I'd have to use
the latest RAD tools. Personally I prefer doeconstructing wrappers to
get clean, small, quick code.

| > An example of what seems like brittleness: I now have a behavior where
| > the Personalization applet opens to the blue box with the white gear
| > and hangs there, for maybe a minute or two. Running Procmon I don't
| > see anything suspicious going on during that time. And this is on a
| > brand new system. It started happening with no action I'm aware of on
| > my part. I did put in a new Desktop picture, but that was before the
| > hang started happening. It's not a big problem, but it is mysterious
| > and gives me pause about depending on Win10.
| | Same happens when starting Win10 in its Safe Mode?
| Why would I boot into safe mode for something like that?

| Whose video hardware are you using: the integral Intel CPU video, or a
| video card in a daughercard slot? Did you match the video driver to the
| hardware source? Was the "new system" a pre-built, or did you build it?

I built it.

Intel 1-5 12400 6-Core Boxed Processor with fan $150
MSI B760M-P Pro Intel LGA 1700 microATX Motherboard $100
G.Skill Ripjaws V 16GB (2 x 8GB) DDR4-3200 PC4-25600 CL16 Dual Channel RAM
$40
Thermaltake Smart Series 700 Watt 80 Plus $60

I had a case and SATA SSDs. But I also bought an M2 SSD,
since the board accommodates it.

The bluescreen tool is interesting. I ran it. It found
a memory dump from the crash and says it was set off
by a realtek (audio) driver. I'm not sure that really tells
me anything.

The only option was a 256 KB dump file, but the bluescreen
reader found whatever was there, anyway.

Newyana2 <Newyana2@invalid.nospam> wrote:

> "VanguardLH" <V@nguard.LH> wrote
>
>| Configure Windows to /not/ reboot on a BSOD.
>
> Yes. See KenW's post above. I didn't know it was an optional
> setting, but now I do and I disabled the rebooting. I'm also going
> to disable the memory dump. Windows wrote a 700MB file!
> It appears to be just raw memory data -- nothing that I have
> the expertise to make sense of.

Max size for a mini (small) dump is 64KB (*). Obviously doesn't include
an image of all of system memory as do the other dump sizes. Just
enable the option to write dumps, and select a small dump size. That's
all you care about as a user. The following describes what a mini dump
records:

https://learn.microsoft.com/en-us/troubleshoot/windows-client/performance/read-small-memory-dump-file

(*) Looks like 64KB used to be the minidump file size. Now it's 256KB.

If you're getting 700MB dump file sizes, you aren't configured to save
minidumps. The only time a larger dump file is needed is when an MS OS
programmer asks you to save a large dump file. They need a copy of
system memory to determine the state of the OS at the time of crash.
They'll use something, like WinDbg, to gain more insight of a crash.

https://www.windowscentral.com/how-open-and-analyze-dump-error-files-windows-10

I only save mini-dumps, and have only used Nirsoft's BlueScreen View to
get some insight on the cause of the crash.

> The thing is, though, that I just built the computer 2 weeks ago.
> It's hard to click in RAM sticks without proper contact, but it's possible
> that was the problem.

Ah, true. Thermal creep wouldn't apply, but oxide on the contacts (on
memory module or inside mobo slot), misalignment, or incomplete
insertion could cause similar problems.

Example: Had a guy who built his own system from parts, but couldn't get
past the beep code indicating a video problem. He had replace a bunched
of parts, but no fix. Contact the mobo maker, and was getting a new
mobo exchange. Before that, he mentioned (threatened) he was bring his
computer into work to swap with parts in Alpha Test machines. NO WAY!
Told him I'd look at his setup. Opened it up, removed the video card,
inspected the mobo slot for damage, replace the video card, voila,
computer booted up. He had an AGP video card which has 2 sets of
connectors: a top row with shorter foils and a bottom row with longer
foils. There are 2 rows of connectors in an AGP slot. Some users only
push hard enough to get the bottom row to engage, and the top row is
barely touching. Have to push harder to get past the "indent" to get
the top row engaged with the slot connectors. Fixed his 6-month old
problem in under 10 minutes. I told him what was the problem: he didn't
push the card all the way into the slot. All I had to do thereafter
when we'd pass each other in the hallways was to grin, and he'd say
"Shut up".

Example: I've seen poorly designed cases where the backplates for
daughtercards was too low, or, conversely, the mobo mount was too high
or the standoffs too long. The tang of the backplate had to be bent
down; else, screwing down the tang meant the other end of the card got
torqued upward, so upward in the slot for poor connection, or no
connection of some pins.

I don't trust pre-builts, so I build my own desktops. With pre-builts,
I inspect to check for faults in production. When we built are own for
the Alpha lab, no such problems. Later, for our QA group, we got
mandated to use Dell pre-builts (think it had to do with leasing and
removing old inventory), and then we had problems unless we did our own
inspection (and then found the same model could have different hardware,
like we'd order 20, and 5 had different a different hardware config
which affected our testing: Dell sells on specs, not hardware
inventory).

> One unusual thing with Memtest: It said there are 12 cores, with 6
> active. Is it possible that I actually bought a much higher powered
> CPU and they've crippled it to sell it down-market?

No, you got a 6-core CPU with hypertheading enabled. Think of
hyperthreading like the interpolated resolution you see advertised for
scanners to bloat their specs.

https://en.wikipedia.org/wiki/Hyper-threading

I forget for which malware (maybe Spectre), but it utilized a timing
defect in hyperthreading regarding speculative execution.

https://en.wikipedia.org/wiki/Hyper-threading
https://en.wikipedia.org/wiki/Speculative_execution
https://en.wikipedia.org/wiki/Transient_execution_CPU_vulnerability

Until Microsoft+Intel came up with a fix (still could occur with AMD
CPUs, but their fix came later), the suggestion was to disable CPU
hyperthreading in a BIOS setting. Some games/gamers also recommend
disabling hyperthreading since it can actually slow game performance. I
haven't run across performance loss with HT enabled, so I leave it that
way.

https://hardforum.com/threads/does-keeping-hyper-threading-enabled-make-any-sense-for-home-systems-especially-on-al-rl-cpus.2026951/
https://www.keysight.com/blogs/en/keys/culture/2023/05/03/keysight-connect-all-about-cpu-hyperthreading

If you want to see more info about your CPU, use the CPU-z utility.
It'll show you the number of real CPU cores ("cores") and the total of
[hyper]thread cores ("threads"). It'll also show the SPD timings read
from the memory modules for you to check if the BIOS settings are
overclocked.

If you are not interested in the miniscule power savings of CPU parking,
you might want to disable that "feature".

https://bitsum.com/parkcontrol/

I also tried their Process Lasso for awhile, but didn't find it to do
much regarding performance in my setups. Those are at a level of
tweaking that I don't bother with, anymore.

I configure my power options (powercfg.cpl) in my desktop PC which is on
24x7 to maximum (high performance mode), and even went into that mode's
advanced settings to further disable power savings. I want it always
ready, and never in a low-power or hybrid hibernate mode. The only
thing that slows my access upon my return is getting past the
password-protected screen saver (which is always there regardless of
power saving settings). I also disable all the glitzy crap that slows
down the UI (sysdm.msc, Advanced tab, Performance, select Custom). I
have about half of all that glitz is disabled. Currently disabled are:

Animate controls and elements inside windows
Animate windows when minimizing and maximizing
Animations in the taskbar
Fade or slide menus into view
Fade or slide Tooltips into view
Fade out menu items after clicking
Save taskbar thumbnail previews
Slide open combo boxes
Smooth-scroll list boxes

If you select "Adjust for best performance", all those glitz features
are disabled.

> Nothing in the event log. At the time it crashed, the copy of the disk
> image was about halfway done, I was running FF, and I was taking a
> file out of a ZIP to read. So, not a lot of strain.

Hopefully it was the memory fault causing the problem. Whether or not
reseating the memory module continues to work remains to be seen.
Fingers crossed both hands.

On 2/12/2024 8:29 AM, Newyana2 wrote:
> "Paul" <nospam@needed.invalid> wrote
>
> |
> | Normally, I would suggest a memtest floppy as a starting point.
> | But the bottom of this post, shows a candidate source of grief.
> | Likely no need to test RAM now. I would be using my portable
> | USB to SD stick right about now.
> |
>
> I think the RPi issue is a red herring. I was using a Windows
> program designed to work with a USB SD adapter. After dealing
> with the RAM issue I had no trouble reading the Pi SD and writing
> the disk image to two other SDs for backup.
>
> As I explained to Rudy, Memtest86 told me I had mucho RAM
> errors. (As in "F0000000 Expected F0000020")
> I took out the second stick and retested. Fine. I carefully
> put back the second stick. Fine. So far, so good. So that sounds
> like a loos stick, right? But I'm still a bit nervous, as no problems
> arose during extensive install and config operations in setting up
> the system. I've been working on it quite a bit for 2 weeks now,
> after having built it. (MSI B760M-P Pro motherboard. G.Skill Ripjaws
> dual channel pair of 8 GB DDR4-3200 RAM.)
>
> Any thoughts on that scenario?
>
>

On a four slot motherboard, you can test two sticks in single channel mode.

channel0 channel1
| |
white white <=== the color of the slots, is to emphasize how to do dual channel
| |
black black

To test RAM single channel,
you would do it like this.

First test case:

channel0 channel1
| |
X stick#1 High_memory
| |
X stick#2 Low_memory

Second test case:

channel0 channel1
| |
X stick#2 High_memory
| |
X stick#1 Low_memory

This exposes different areas at the top and bottom of each
stick, and comes closer to 100% coverage. (The E801 reserved space
cannot be tested, but flipping the sticks, the E801 spaces don't
overlap.)

You still have to use your thinking cap, with regard
to the failure addresses returned when testing like that.
Some useful info is still missing.

*******

When testing a MicroATX, the two DIMM channel layout can be like this.

channel0 channel1
| |
Black Black

Since there is no really good, single channel, test there,
you're limited to testing a stick at a time, for unambiguous
test results. The E801 addresses are entirely missed, but
the test coverage is still a very very large percentage.

channel0 channel1
| |
Black stick#1 <=== testing this way is better than nothing

Windows has a memory tester, but it will be storing the kernel
in about 300MB of that RAM, which means the coverage isn't quite the
same as with Memtest. The Windows test algo can be better (I've had
Windows detect trouble, that memtest could not see). But the ability to test
the entire DIMM, is limited for a Windows memory tester operation.

This means, for cases where you suspect memory is the problem,
and yet memtest cannot find it (even using two sticks in single
channel mode), you try the Windows test anyway. Because it does
have some useful attributes. It's not a total dog.

*******

https://download.msi.com/archive/mnu_exe/mb/PROB760M-PDDR4.pdf

Page 25 A1-A2 B1-B2
\---/
|
+------- Use these slots for single-channel two-stick test

Normally, for dual channel, and for the first two DIMMs, you
would use A2 and B2. (Like if you were preparing your "gaming machine"
for best performance, it would be A2 and B2.) But for the Memtest,
I'd use B1-B2, then swap the sticks and retest.

The purpose of this test, is "thoroughness".

For fault isolation to the nearest stick, you can still test the sticks one at a time
in B2 slot. But if the fault will not show in the memtest screen,
while testing with just a stick in B2, you may be "stuck" with the
test results obtained from your B1-B2 testing.

I've had failures that only show when all four sticks are installed,
and no other test shows a problem. The end result ? All four sticks
replaced, and problem solved :-/ That was on my WinXP machine (which
is now dead). It appears the Southbridge blew, rather than the
Northbridge that has the memory controller on it.

On your new machine, the CPU has the memory controller, the motherboard
only has a weeny Southbridge on it (they do not have the technical
content they once had). Some AMD processors have a silicon area
referred to as SOC or System-On-a-Chip, and quite a few I/O functions
can actually be on the main CPU. The reason a Southbridge exists,
is for the whizzy Asmedia USB logic blocks in it. Some day, that
might be USB4, when it's ready.

Paul

Newyana2 <Newyana2@invalid.nospam> wrote:

> "VanguardLH" <V@nguard.LH> wrote
>
>| Did you let Windows Update (WU) install the hardware drivers, or did you
>| get the drivers from the hardware manufacturers own web site?
>
> V! What do you take me for? :) I always go to the manufacturer.
> This is the first time the MB didn't come with a disk, but I found
> the drivers online without trouble.

I don't compile and retain biographies or histories of attributes or
expertise of posters in Usenet. I monitor 52 newsgroups. Some where I
lurk, and some where I participate. Way too many posters to keep track
of. I may recognise a nym, but that doesn't mean I have their expertise
memorized.

> Why would I boot into safe mode for something like that?

Non-critical services and startup programs can cause interference or
conflicts.

Since the only time you encountered a BSOD, so far, is when using the
Win32 Disk Imager program, maybe the imager program has a flaw. Its
last update was back in June 2018.

A Github project is a rewrite of the Sourceforge one you're using is at:

https://github.com/znone/Win32DiskImager
This project was rewritten from Win32 Disk Imager
(http://sourceforge.net/projects/win32diskimager/). Both interface and
function are basically the same. The difference is that the project
was written using Win32 native technology, while the original project
was written in QT.

Unfortunately you have to compile to code for the rewrite version
(https://github.com/znone/Win32DiskImager/tree/master/DiskImager). No
already compiled binaries to download.

I didn't see anything listed under Open Issues for the rewrite project.
Looks like the rewrite Github project has been around for 5 years.
There are still many open bugs for the original one you used from:

https://sourceforge.net/p/win32diskimager/tickets/

Plus its Sourceforge project home page mentions some other problems,
like:

Warning: Issues have been reported when using to write to USB Floppy
drives (and occasionally other USB devices, although very rare). While
this has been fixed in v1.0, it is highly recommended that before an
image is written to a device, the user should do a Read to a temporary
file first. If this fails, please report the failure along with your
system information.

My guess is "read to a temporary file first" means making sure the
imager program can, at least, read a file from the USB media as a test
that the program can access the USB media which hints the program can
likely write to it.

> I built it.
>
> Intel 1-5 12400 6-Core Boxed Processor with fan $150
> MSI B760M-P Pro Intel LGA 1700 microATX Motherboard $100
> G.Skill Ripjaws V 16GB (2 x 8GB) DDR4-3200 PC4-25600 CL16 Dual Channel RAM
> $40
> Thermaltake Smart Series 700 Watt 80 Plus $60

Mostly interested if you got driver updates via WU, or you got them from
the hardware maker. You said the latter, but did you disable hardware
updates in WU? I've seen users mentioned they installed the mfr driver
only to find later the WU driver got installed.

> I had a case and SATA SSDs. But I also bought an M2 SSD,
> since the board accommodates it.

Just be aware that often using the m.2 slot in the mobo will disable
some of its SATA ports. The manual should warn you about that.