There was apparently a power drop at my home today, I came back and found my computer sitting at the KDE Desktop login screen. Once the computer finished logging into desktop, BOINC auto-started and I immediately saw all my Einstein tasks reporting "GPU Missing". I've had this message before, and upgrading my nVidia drivers fixed it -- but now I'm on Debian 12.11 (previously Kubuntu 22.04) and my nVidia isn't a Snap -- and appears to be the latest version (535). This seems to be a false report, because the nVidia settings app opens and finds the GPU and driver.

I've restarted the computer, then tried complete shutdown and left off for most of a minute, no change.

Potentially related, MilkyWay reports "Running 4 CPUs" -- but I have eight cores in my AMD Fx8350 and MilkyWay normally reports "8 CPUs" on this system.

AMD Fx8350, 32 GB RAM, plenty of free space (BOINC has its own partition with tens of GB free and allowed to use up to 20 GB). RTx 2070 with nVidia 535 driver. Debian 12.11 kept updated (haven't found the time and nerve to upgrade to Debian 13 yet).

Suggestions?

One thing to be aware of is that MilkyWay sometimes will send out tasks that use less than the normal number of cores ( a feature designed to confuse). That said your other problems may be related to the apparent power outage.
First, check the number of cores your operating system is seeing - as you say it should be 8 provided all is well. If not then it is possible that the BIOS settings have been corrupted, you will have to shut down, then do a restart and access the BIOS (probably either "F2" or "del" during the start and make sure that all is well - it's many years since I had an FX8350 based computer so I can't point you to the right section in the BIOS menu to find the settings, from memory it is possible to over ride the "ignore suspect cores" option (which might be "use all cores"). Hopefully this will put things back the way they should be.
Hopefully you will be able to get all eight cores back working and this isn't the signs of a CPU that is starting to fail.

When the GPU is 'missing' does restarting the BOINC client solve the issue?

I've had a Linux PC in the past where BOINC started before everything else loaded during a reboot so it thought there was no GPU. Although I think it was an AMD GPU at the time.

If boinc client restart fixes this issue with the drivers - then it's definitely an issue with the boinc client being loaded before the drivers are initialized properly.
Usually boinc client is started by the systemd, so fixing the boinc-client.service file by adding a delay will fix the issue.
If this is an issue related to the drivers being installed as a snap - then this is a different issue, and you should report it to the GitHub.
But please, before reporting it be sure that this is not the first case I described above.

---

BOINC maintainer.
For any insight, check my BOINC Development Blog.