Mount /tmp as tmpfs

[BitJam]

Thank you for your considered reply fehlix! I'm not sure how much more proof you need. In order to wear out those 2013 drives, they had to write to them nearly continuously for about a year, some much longer than that. The duty cycle of writes in normal use is vastly smaller that that. You could reinstall MX every day of the year and only do a small fraction of that writing. What's it take? Five minutes on a fast drive? Call it 5G. That's about .01% of the daily writes used during the torture test. Likewise, updating access time stamps once a day is much less writing than that even assuming every file is read from. The size of the data is much greater than the size of the metadata.

The SMART system tells you the total bytes written, at least on some SSDs. I suppose someone could try to do a measurement comparing the bytes written when noatime is used versus the bytes written when relatime is but I think you would have to measure for months to test this under normal operation and I imagine any difference might still be lost in the noise. Even if it increased the writing by 10% (which I doubt) this is literally a small fraction of a small fraction of the writes needed to kill a drive.

There is no question that using atime can hurt performance and probably longevity. The relatime option became the default back in 2007. Here is part of the story behind it: Once upon atime. So there was a time, over a decade ago when using noatime on flash drives made sense.

(Too Many) Notes
It is my opinion that the noatime suggestion and the suggestion for no journaling dates back to the very early days of usb flash sticks. This may have been before relatime was a viable option (or became the default). Years ago, I searched and searched for any hard data indicating these options increased usb stick longevity. I could not find any. I think part of it was people were pushing every button they could push. But I think the theories had some validity way back when if the very early flash sticks had no wear levelling. In that case, frequent writes to the same location (directory inodes and the journal area) could kill a stick before its time. Then, unfortunately, these suggestions got propagated and amplified over time by the internet echo chamber even as technology developed and they became less and less relevant.

Back around 2009, anticapitalista and I decided to focus a substantial part of our development effort on live-usb technology which led to the Most Extensive Live-usb on the Planet! Even back in 2009 it was clear to us that live-usbs were better than live-dvds and live-cds. It was also clear that the usb flash drive technology was going to improve drastically in the coming years leaving cds and dvds in the dust. This happened Running live on usb-3 is amazingly fast. With fast hardware I can install and boot a Frugal system with persistence in less than a minute. This long-term live-usb project is why I researched the "noatime" and "no journaling" claims years ago.

But those old rumors are hard to kill. That is one of the reasons why live-usb-maker always creates a journaling filesystem. Some people still cling to those old rumors which, even if they were applicable over a decade ago, are certainly not applicable now. If you go through the MX forums, you will see that most of the non-recoverable live-usb data loss was associated with using a non-journaling filesystem.

[fehlix]

.. I'm not sure how much more proof you need.
.. But those old rumors are hard to kill.

This last sentence is I'm all about. And I'm so thankful you summarized all your clear thoughts in such consize manner. This is more than enough to have a well educated and well proven statement from with MX/antiX community and development team to go further.
fehlix