The basic symptom is:
Running system gives "black screen"
HID devices are disabled. Plugging in a new HID device, it is ignored on purpose.
NIC is disabled, LEDs go off on it.
The basic symptom is:
Running system gives "black screen"
HID devices are disabled. Plugging in a new HID device, it is ignored on purpose.
With no keyboard, you cannot ctrl-alt-delete or any other input value.
NIC is disabled, LEDs go off on it.
USB ports *do* have power, but some detection constraint prevents usage.
In the past, no EventVwr.msc log entry, was available. But some
things can be logged, like Watchdogs can be logged.
At one point, while working on it, it became SO BAD, I couldn't even type in the PIN before the computer black screened. It was not inclined in that case, to wait 60 seconds before dropping. Previously it would only do these about once a week. There could be periods where it stopped happening entirely,
but it's been a dumpster fire recently. If I log in with my second profile, it is all "sweetness and light" and the frequency of the event and the predictability are a lot lower.
On 10/19/2026 3:17 AM, Paul wrote:
The basic symptom is:
Running system gives "black screen"
HID devices are disabled. Plugging in a new HID device, it is ignored on purpose.
With no keyboard, you cannot ctrl-alt-delete or any other input value.
NIC is disabled, LEDs go off on it.
USB ports *do* have power, but some detection constraint prevents usage.
In the past, no EventVwr.msc log entry, was available. But some
things can be logged, like Watchdogs can be logged.
At one point, while working on it, it became SO BAD, I couldn't even type in >> the PIN before the computer black screened. It was not inclined in that case,
to wait 60 seconds before dropping. Previously it would only do these about >> once a week. There could be periods where it stopped happening entirely,
but it's been a dumpster fire recently. If I log in with my second profile, >> it is all "sweetness and light" and the frequency of the event and the
predictability are a lot lower.
Are you indicating that the exact same issue occurs in the second profile but after the ability to logon(enter pin, access desktop and account settings, etc.) before locking up.
If so, please define what you meant by frequency(minutes, hours)?
On Mon, 10/19/2026 2:57 PM, ....winston wrote:
On 10/19/2026 3:17 AM, Paul wrote:
The basic symptom is:
Running system gives "black screen"
HID devices are disabled. Plugging in a new HID device, it is ignored on purpose.
With no keyboard, you cannot ctrl-alt-delete or any other input value.
NIC is disabled, LEDs go off on it.
USB ports *do* have power, but some detection constraint prevents usage. >>> In the past, no EventVwr.msc log entry, was available. But some
things can be logged, like Watchdogs can be logged.
At one point, while working on it, it became SO BAD, I couldn't even type in
the PIN before the computer black screened. It was not inclined in that case,
to wait 60 seconds before dropping. Previously it would only do these about >>> once a week. There could be periods where it stopped happening entirely, >>> but it's been a dumpster fire recently. If I log in with my second profile, >>> it is all "sweetness and light" and the frequency of the event and the
predictability are a lot lower.
Are you indicating that the exact same issue occurs in the second profile but after the ability to logon(enter pin, access desktop and account settings, etc.) before locking up.
If so, please define what you meant by frequency(minutes, hours)?
It used to happen once a week, at random.
Recently, it changed to being quite reproducible.
Boot the computer, use my Daily Driver profile, login,
after around 60 second, screen would go black, all HID disabled,
USB ports *do* have power, but device LED only "blinks" as
the OS perhaps reads the config space.
If I log in with a second profile ("Bullwinkle"), it would
stay up for longer.
Figuring this was the containerized Nvidia driver doing this,
I turned on the iGPU as well as leaving the NVidia card installed.
Now, on the Daily Driver profile, I couldn't even type in the PIN,
it dropped so fast. I would not have been able to select "Bullwinkle"
profile either.
It was during this super-fast in-the-middle-of-login failure,
the computer started doing a "diagnose" with the spinning icon,
and it must have changed something. Looking in the SRT trail folder,
there is lots of activity, but nothing that points to a specific cure.
I do not believe, for example, that any sort of DISM command can
finish in seven seconds. Maybe if a health issue was logged previously,
it could harvest that.
*******
When the machine was assembled by me originally, it was flawless.
I ran using just the iGPU for several months. Then, one day, it started...
The RealTek NIC would drop out, around 30 seconds after the desktop appeared. This was reproducible, each time, the same time constant. Changing driver
did not help. (Today, the RealTek NIC is containerized as I understand it.)
I replaced the RealTek NIC with an Intel NIC card. Good so far.
Then, later, the iGPU started doing the black screen thing (Vega 8
on the Zen3 5700G 8 core). I installed a $200 video card from the
computer store, a GTX1650. I turned off the iGPU. Things were fine for
a while. Some months, I could go the whole month, without an event.
Other times, it would be about once a week.
Just recently, the condition became "no longer random". taskschd.msc
no longer has any NVidia tasks (there used to be four of them
related to NVidia telemetry). Can't blame it on that then. Something
else is triggering the NVidia behavior (I think something is attempting
to contact the driver and is not succeeding). I changed driver back
to an older one, didn't help. It's hard to tell what Nvidia software
is installed, as there is nothing in Apps, nothing in Programs and Features. And there is no C:\NVidia folder, where sometimes installers unpack themselves.
The C:\AMD folder still exists for the AMD video driver of the iGPU (Vega8).
By "no longer random", with just the GTX1650, the screen would go black
about 60 seconds after entering the PIN. That's the GTX1650, with the
iGPU switched off. Then, after trying a few more things, I couldn't
even change the PIN (or select "Bullwinkle" secondary profile). But
when that happened, finally windows used the "Diagnose" spinner. And
somehow, I'm running and the Reliability Monitor is not showing any
"NVidia Container" entries for today so far. The 4K monitor is being
driven by the GTX1650 HDMI out.
If I do memtest, it passes.
If I run Linux, it's fine. I've never seen a "similar symptom" while testing Linux on this machine. I could "make a kernel panic" in the past,
via overclocking one of my older single-core machines :-) You can still have boot failures on poorly prepared distros. The worst distros are a sight
to behold (desktop won't start, mouse stops working). But a Ubuntu
or a Linux Mint, is generally well prepared and works fine. And no
symptoms that "smell of hardware level trouble".
I have debugged other problems, via comparison analysis. Run Windows,
get a crash in 20 seconds. Run Linux, get a crash in 20 seconds
(Linux twitching before it dies). This is part of a hardware issue
with the 440BX chipset on Intel. Originally I was blaming Windows,
but when I got the Linux result, then I knew it was hardware and
the address map for memory was somehow not right. Installing 768MB
of memory on 440BX, spells instant trouble (256MB+256MB+256MB).
Slightly less than 768MB, you could run Prime95 for 16 hours with no problem at all (256MB+256MB+128MB+64MB). Very stable, and using
four sticks of RAM too on a single channel bus.
Summary: While I'm running on my Daily Driver profile right now,
I really don't know what will happen next. And with no evidence
of what has changed to stop it, I won't know what to do or where
to go the next time. You'll notice I'm not volunteering to do
a Clean Install (yet).
Paul
I was not going to suggest a Clean Install( yet :) ).
ÿYou've mentioned quite a bit about the Nvidia card
ÿ- Does the Nvidia graphics card connect via PCI Express 3 x16 slot or via a single or double 6-pin power connection direct from the power supply?
ÿ- Have you considered removing that card from the device to only run on the onboard video?
ÿÿ Or before complete removal, if possible insert in another mobo slot or power connector( Note: It's been a while since I've seen a power supply with a single 6-pin connector though some double connectors allow removal of the 2 pin for a stand-alone single 6-pin)
ÿI'd also look at removing the CPU fan/cooler to determine that heat dissipation is not blocked and also if the fan is operating at intended rpm for cooling(on old mobos I've seen a few top of cpu cooler fans on old devices with a bad bearing operating well below 200 RPM(and a lot of fuzz and dust blocking heat dissipation).
It's been a while since I've seen or heard about Pentium I/II 440BX(which also indicates an older mobo).
Finally, does the device lose user control if booting from a Windows USB drive and running through some non-destructive tasks(command mode, diskpart, troubeshooting)?
| Sysop: | Jacob Catayoc |
|---|---|
| Location: | Pasay City, Metro Manila, Philippines |
| Users: | 5 |
| Nodes: | 4 (0 / 4) |
| Uptime: | 498678:37:06 |
| Calls: | 212 |
| Calls today: | 3 |
| Files: | 912 |
| D/L today: |
31 files (22,593K bytes) |
| Messages: | 69,880 |