So I've noticed a few of these happening, maybe one per day, and they have an error code of:

(unknown error) - exit code -1073741819 (0xc0000005)

I did have a poorly seated memory module back on the 13/14th of March which caused some errors, but that's fine after a reseating and it passes memory diags correctly.

But I still see a few of these errors. The memory is being run at spec, the CPU is at 50C. I don't see any stability issues with anything else.

So are these a source of concern? Or could they have been naturally occuring and I'm only noticing them now after the memory issue drew my attention to them?

http://www.primegrid.com/result.php?resultid=796048243
http://www.primegrid.com/result.php?resultid=796046714
http://www.primegrid.com/result.php?resultid=795498465

If you search these forums for that error code, you'll discover:

A) That code means "access violation", which can be caused by almost anything.

B) It's been known to occur on many different apps.

C) Since almost anything can cause this problem, it's not surprising that there are posts about it going back almost to the beginning of PrimeGrid.

You said that reseating the memory cleared up some problems. Are there any smokers in your household? Is the computer relatively close to where food is prepared? In both cases, smoke (either from tobacco or cooking) could be contaminating the connectors on the memory. An old trick is to clean the gold memory contacts using a pencil eraser. Even in an environment that isn't smokey, you would be surprised how much grime can collect on those contacts.
____________
My lucky number is 75898⁵²⁴²⁸⁸+1

Is this the i7-5960X? If so, is it overclocked at all?

It is overclocked from 3.0GHz to 3.5Ghz via an XMP profile, and has never shown any stability issues. A nice big Noctua fan ensures it's nicely cooled. But the RAM is at it's rated 3GHz spec.

None of the other various applications have exhibited this though.

There's no smokers, debris or other ventilation issues, 12 case fans ensure clean air flow.

So, I've fixed this.

And it was weird.

The machine was also bluescreening randomly, sometimes once every 36 hours, sometimes only lasting 1 hour.

Checking the .dmp files showed random causes and random drivers associated. Since this seemed coincided with the Windows 10 creator update, I tried reinstalling drivers, and then a clean install after rollbacks botched the install.

It would pass Memtest86, Windows Memory Diag, and HCI Memtest. Which seems to suggest the RAM is okay. But running Prime95 with small FFT is fine, suggesting the CPU is fine, but the blend test would failed randomly after only a few minutes, which again would suggest the RAM is bad.

When that didn't fix it, I settled down for the slow chore of checking each stick one by one.

But as I started through the RAM sticks, I noted something odd in the UEFI/BIOS, one of the video cards had dropped from x16 PCI-E lanes to only 8. The only change I had made in terms of the PCI lanes was to move the soundcard from one slot to another so as not to block a USB connector.

Once the card was moved back, lo and behold, all the errors are gone.

I can't begin to guess how that works, or how something as innocent as a soundcard in a differing slot can compromise a machine like that.

So it now passes Prime95 blend, and has already a few hundred units through without a single issue, inconclusive, error'd or otherwise. As if to underscore this it even produced my 8th ever prime today as well.

No, the BIOS wasn't touched, either in terms of updating, or making any changes. It's been the same for months. Plus there's nothing in Win10 that would set the graphics card back to 8x PCIE lanes, since that happens long before any OS is touched.

So this turned out to have a very different and unpleasant root cause.

At some point the processor socket suffered two bent pins. I don't know how, I can only imagine the heatsink took a knock or similar, and two of them didn't take too kindly to the flexing. These pins were evidently involved with the memory controller for the banks to the right of the socket.

Moving the sound card just exacerbated the poor contacts further.

There's no RMA possible, no repair place will touch this kind of error, and I couldn't get the pins back myself either, so I've had to replace the board.

Thankfully nothing else was damaged, but it's more than a bit galling to be out €400 on a board over such a small and inadvertent thing.

One was bent over and twisted somewhat into a second, also pushing the second. While the second one pushed back okay, the first one wouldn't go back at all. Zooming in with a camera on it, it looks like the contact point on the end of it is melted onto the second pin.

Checking google shows I'm far from alone in this experience sadly.