On Wed, 25 Mar 2020 03:48:22 -0400, Paul wrote:

At one time, and especially with AMD, the slot population
rules were more restricted. Dual channel mode only worked
in "128 bit mode", with the two sticks in equivalent slots
on each channel. I don't think AMD needs that any more.

Hi Paul,

Thanks for your always purposefully helpful adult advice, where I'm not
exactly sure what a "channel" means since I haven't touched a memory stick
in more than a decade; but I'm currently running with two 4GB memory cards
set in the two black slots at left (which seem to be Memory Bank 3 & 4).
Full https://i.postimg.cc/y6bKSHPB/bsod23.jpg
Empty https://i.postimg.cc/L4ZfrHh0/bsod22.jpg
Banks 3 & 4 https://i.postimg.cc/J0G9Qp8t/bsod28.jpg
Boot confirmation https://i.postimg.cc/bY4t9dHw/bsod29.jpg

It seems to me that, even though the memory has consistently tested as
being OK, there are still a LOT of memory tests I can run (based on my
prior post where I haven't run all the memory test tools yet).

While nothing seems repeatable, one oddity is that when I immediately
reboot after a BSOD, it consistently provides a bunch of BDODs, but, if I
pull the power cord and wait for the LED on the motherboard to die out, it
seems to come up clean... so ... when it gets a BSOD... something... but
what... is "saved" somewhere as an error condition.

You can test two sticks in single channel mode, when using
the memtest.org utility, as a means to get 100% memory coverage.

I'll run the memory test again, with only the two sticks in the black slots
at left which appear to be memory banks 3 and 4.

But for a quick test, I'd just try two sticks in dual channel,
for an immediate reward.

I admit I don't really understand what you mean by single channel and dual
channel. I'll need to look that up to understand what you're trying to
suggest I try to do.

If you want to turn on XMP or its AMD equivalent, sure, you can.

I might do that later, but first I need to test the memory again, with only
the two left-most "black" (memory banks 3 & 4) and then the two right most
"blue" banks later (memory banks 1 & 2).

It depends on whether you want to understand what mistake you
made with the original config, or, you just want it fixed.

The only thing that I know happened is it first BSOD'd when I was cleaning
out the fur inside the fan on the CPU with bottled air and I was cleaing
out the fan on the NVIDIA card while the machine was running (which, in
hindsight, was a dumb idea).

There are some NVidia chipsets, where turning the memory clock down
is really the only practical working solution. Most motherboards
there are multiple knobs worth twisting.

A few days ago I removed the Nvidia graphics card, hoping that was the
problem (as the GPU fan was making noise anyway), but the problem has
persisted even with the Nvida graphics card wholly removed:
https://i.postimg.cc/SK6J1Z4c/bsod30.jpg

At the moment, I'm testing with two memory cards removed (bank 1 and bank 2
are both removed) and with the DVD/CD optical drive removed, and with the
Nvidia graphics card removed.

My next test will be to see if the AMD-based HP machine will boot with just
one memory card, which then I can test, one by one, each of the four memory
cards, to see if it's the memory (or something else).

Thanks again for your always purposefully helpful adult intent on Usenet.
--
Too bad this thread isn't archived by Google groups for future reference.
https://alt.comp.os.windows-10.narkive.com/oL7PTNKu/windows-10-bsod-indicates-a-hardware-problem-but-what-hardware-is-the-problem