The Solid State Drive (SSD) has been in the system for over a week now, so here’s an update on it:
It’s FAST!
Okay, that’s not news, but I felt compelled to say it anyway. 😉
After letting the Linux VM run for several days with memory-hog programs like Firefox (with about a dozen pages open) going, and developing code for at least a few hours each day, free memory was getting pretty low. When that happened previously, I’d start getting massive delays when I tried to compile things, with tools indicating that the system was waiting for the hard drive (“IOWait”). As I’d hoped but didn’t dare expect, those delays are GONE! Every compile is equally fast now, whether the first one after a reboot (which I’d previously clocked at taking nearly four minutes) or the twentieth, with several gigabytes free or only one! I can barely see any IOWait delays at all, in action or even on the monitoring program!
I’d REALLY like to move the system drive of the Windows VM over too, but as mentioned before, that would take a reinstall of the OS and a lot of time that could be much better spent writing code. The Windows VM doesn’t seem to suffer nearly as much from low-memory-induced compile delays anyway, so as nice as that would be, it can wait a while longer.
I can definitely see why people recommend SSDs for their speed, despite their well-documented longevity problems. Now if this one will only last for a few years…
As a very minor update: I was doing some cleanup work on the system today and had the System Information program open. For giggles, I called up the Serial-ATA screen and looked at the new SSD. To my surprise and consternation, it was reporting that the drive was not using the TRIM command!
Since that command is important for drive longevity and maintaining its speed over time, it couldn’t be left that way. It took a bit to dig up a page that described the problem thoroughly, but persistence paid off. It seems that Apple decided to only enable TRIM for its own drives, the ones that it preinstalls into systems. Drives installed by the user are given the cold shoulder.
Fortunately that page also describes the solution, which I tried. It seems to have been successful. I don’t notice any major difference in speed, but that’s kind of the point. 😉