Virtual memory isn’t swap, it is a mechanism that allows the operating system to give processes a view of memory that is almost completely decoupled from real physical memory and other processes. For example some programs require their code and data to be placed at exact memory locations in order to work - virtual memory allows you to run as many of these programs as you wish, because one process’s address 0x1000 has nothing to do with another one’s 0x1000, unless they set it up as shared memory (but even the same chunk of shared memory might be mapped to different addresses in the processes that share it).

Swapping is a cool trick that you can do with virtual memory, though. Basically you store a piece of memory somewhere outside the physical memory, and then make the address invalid in virtual memory. When the process tries to access it, it will crash. The OS will be notified of the crash, see that it was due to the process trying to access swapped out memory, load the chunk back from disk (maybe to a different physical location), update the virtual memory to correctly point to this chunk, and restart the crashed process from the instruction that caused the crash. So from the point of view of the process, nothing went wrong at all, except that one instruction took a very long time to execute.

Also, isn’t it harmful to SSDs?

Swapping doesn’t do enough writes to matter, unless your system is running really low on RAM.

xfwm is XFCE’s window manager, and it’s eating almost 30% of the total system memory, so that’s the prime suspect (I’m not exactly sure how much it interacts with other apps, so it’s possible something else is forcing xfwm to use all that memory, but that is IMHO unlikely).

An ugly “fix” is to log out and log back in (yes, not much better than just rebooting), or you could try to somehow restart xfwm - running xfvm --replace in terminal might work.

Edit: there’s an issue on the Manjaro forums that might be related: https://forum.manjaro.org/t/xfwm4-memory-leak-since-4-20/173910/7

That was my deleted comment, but then I realized the article specifies they are Rome CPUs (Zen 2), while the 4000 series of EPYCs is based on Zen 4

deleted by creator

Potentially the same thing, assuming PCIe 2 x1 provides enough bandwidth.

The thing that’s going to be locked behind the subscription is your ability to watch those files on your NAS through Plex when you’re not in the same network as the Plex server. That’s streaming.

If you only use Plex while at home, you will indeed be unaffected

I don’t think overheating would cause random corruptions (it should throttle down when overheating, and then shut down if the temperature gets too high even when throttled, but there should never be an incorrect result of any computation), and surely the RAM will run at the standard 2133 speed on default settings - OP says they reset the BIOS settings to default between CPU swaps.

A simple rm -rf says hello

Google Maps is US-based whether you use it in the US or anywhere else

Sure, but I don’t see how any of that disproves the current “M$ supremacy” for “normies” - the fact is that people who couldn’t care less about how their computers work will have a much easier time using Windows (and probably macOS) than any Linux distro. You don’t have to worry that some software won’t be available to you because of your choice of the OS, and if you ever have a problem it’s easy to find help.

I haven’t used Windows in a decade on my personal computers, but as long as these two things hold true, it will always be my recommended OS for people who simply don’t care - I’m not going to spend my time doing free IT support for everyone I know and then get blamed everytime something doesn’t work.

Annoying warning keeps showing up at boot -> bring the PC to the nearest computer-literate person, and they’ll fix it. Good luck doing the same if you use Linux.

Obviously you need some redundancy in case the script gets corrupted. 5000 repetitions seems reasonable for such a high quality work

Yeah, stock Google voice recognition also works offline if you download the language model beforehand.

Indeed, try switching your smartphone to airplane mode and see how far your voice commands get you.

Did that (or rather disabled mobile data and WiFi, because airplane mode would still keep the WiFi on), and then I dictated this sentence after the parentheses. So Google’s voice input works offline just fine.

Or do they mean something like a smart assistant? In that case fair, but it’s not like it will work with text input either.

It is true, however, that Google Translate doesn’t do offline voice translation even if the language you’re trying to translate from is downloaded for system-wide voice recognition.

Don’t be ridiculous - this is a lab environment, they can faithfully recreate the suffering as long as the ethics committee doesn’t get notified.

That’s a reasonable per-core size, and it doesn’t make much sense to add all the cores up if your goal is to fit your data within L2 (like in the article)