Somewere in the run up to Slackware64 15.0 release and kernel 5.15.xx,
two odd stability quirks started happening.  On two systems:

Gigabyte X570 AORUS ELITE with AMD Ryzen 9 3900X, Radeon RX 5700 XT
Gigabyte X570 I AORUS PRO WIFI with AMD Ryzen 7 3800X, Radeon RX 590

(Both are XFX brands.)

These systems would lock up at random intervals, with days or weeks of
uptime, and with errors in the logs pointing to X.org and amdgpu kernel
module errors. When th system locked up, the last X display was still up
on the monitor as a static image.  Usually, remote ssh was enough to get
to reboot.  However, the RX 5700 XT system sometimes was completely
locked: no ssh, and no log messages when power cycled. There was even a
lockup (ssh was still available) once overnight when X.org was not
running. Searching the interwebs hinted that something was going on with
PCI-e 4.0 or that the amdgpu kernel module. This was weird, since kernel
5.14.xx was very stable.

I replaced the RX 590 with and Nvidia GTX 560 and that system no longer
locks up in X.org. Perhaps a PCI-e 3.0 vs 4.0 issue? The RX 5700 XT was
running as a PCI-e 4.0 x16. The lockups, of course, only happens when
the system is idle - power management off, no screen saveer etc.

The second issue was a problems with USB 2.0 flash drives. The X570 has
USB 3.0 controllers.  These worked fine until one day not (!) for a
gerneric office store brand USB 2.0 stick. USB 3.x sticks worked as
expected. This led me to an old post:

https://gist.github.com/oakwhiz/1dcec793e5f728556459

that there was a problem with the VIA USB 3.0 controller and the AMD
IOMMU function with some AMD motherboards and firmware. I had just
recently plugged in a Gigabyte IPS monitor with integrated USB 3.0:

Bus 006 Device 004: ID 2109:0817 VIA Labs, Inc. USB3.0 Hub
Bus 005 Device 007: ID 2109:8883 VIA Labs, Inc. USB Billboard Device
Bus 005 Device 006: ID 2109:2817 VIA Labs, Inc. USB2.0 Hub

Per the article, adding

amd_iommu=on iommu=pt iommu=1

to the linux boot append or turning IOMMU off in the BIOS fixed this
issue. I opted to disable IOMMU.

The GPU and sometimes system hard lock still occurs. I turned off
AMD COU cool and quiet mode as well.  Both systems have liquid CPU
cooling.

This lock up seems to be related to the AMD GPU, since the system with
the Nvidia GPU is stable. I'll keep looking. GPU prices are finally
coming down so hopefully I can also try another AMD GPU.