NVMe drive not releasing disk space after failed plot

For mechanical drives, it does rearrange the clusters for a given file. For SSDs it runs trim only.

I would change schedule for daily, so you don’t need to remember it anymore. However, I would also run it manually or through a script (sleep 15 mins or so) to see how it changes your plotting times (basically not much, so it is not really worth it to bother).

Trimming doesn’t touch “clean” cells, only those that are marked “dirty.” And those dirty cells are scheduled for an explicit trim when being written to the next time (thus the slow-down everyone talks about), so should be no harm. At least, this is how I understand it (most likely too simplistic). Although, if anyone has a better understanding about the trim process, it would be nice to get the correct interpretation.