Fragmentation over time and TRIMBefore we get into it, here’s what the Vertex 2 Pro looks like brandy new and unmolested by our testing, with only a single write pass conducted as we do with every new SSD (necessary as some SSD’s “cheat” when brand new and respond to requests faster than they would under normal use):
Read: 215.8 / Write: 242.3
I normally fragment a drive with a bunch of random access on top of our already brutal benchmark suite. I hammered on the Pro for a bit and checked out the results:
Read: 184.5 / Write: 241.5 (!!!)
The Vertex 2 Pro shrugged off any fragmentation effects, at least in the area of writes. You might expect the writes to be that lower line above, but those are the read speeds. Writes are still riding the 240 MB/sec line! It appears the added overhead of the previous random writes only impacted the read speeds, dropping them to 185 MB/sec. If you think about it, this really is not an issue, as it would be very rare for the drive to sequentially read areas that have previously been randomly written to. Lets see what happens after an additional pass, where drives with ‘smarter’ firmware will self-heal on the fly:
Read: 214.3 / Write: 243.6
Read: 214.3 / Write: 243.6
Sequential writes returned the Vertex 2 back to its happy place. This was not the case with the Vertex 1, which required either time (auto-GC) or TRIM to be returned to its expected level of performance.
The Vertex 2 Pro did so well here that I was hesitant to even bother with TRIM testing. Why? It’s so fast it almost doesn’t need it! But hey, lets try it anyway. I performed a few write passes to bring the drive back to non-new status and had Windows 7 issue TRIM across the drive. The result:
Read: 214.9 / Write: 242.2
This is one instance where I’m not really going to complain that TRIM had a negligible impact, mainly because the drive performs so well without it. This is actually a heck of a bonus, as no RAID solutions are currently passing TRIM, so a pair (or more) of these drives running in a RAID should scale nicely and suffer no perceptible long term fragmentation effects.
Moving on, the OCZ Toolbox had a drive wipe feature, so I ran another pass after a drive wipe to see the effect of a fresh (i.e. factory new) drive. It’s worth noting the wipe was *extremely* fast – we’re talking less than 1 second. I partly ran this test just to make sure the tool actually did something 🙂 :
Read: 246.6 / Write: 242.9
This is ‘new out of the box’ performance. Typical use will cause performance to drop back to more realistic figures, but for the Vertex 2 Pro, even ‘normal’ is damn fast!