Storage Consistency: The Adobe Photoshop (Heavy) Trace
Storage Consistency With PCMark 8's Adobe Photoshop (Heavy) Trace
Because there are 18 individual rounds packed with 10 traces each, we need to focus. We'll choose one trace, Adobe Photoshop (Heavy), and keep tabs on it through the entire extended run.
Bandwidth
Most of the data accumulated for Plextor's M5 Pro is based on older firmware. We ran the PCMark tests after flashing to version 1.06 though, and it appears to enjoy much better resiliency. Apparently, Plextor made some solid improvements to the technology it calls TrueSpeed, which includes better algorithms to keep write amplification low and performance high when the drive is out of available blocks for writes. The M6es fare well enough, but the M5 Pro handles PCMark 8 much better.
These drives are already in a steady state by the time PCMark 8 gets to the Photoshop trace in round one, so performance starts low. Already, though, the M5 Pro sustains two times more read and write throughput than the M6es. Once we get to the Recovery Phases, it establishes an even greater advantage. We're currently looking into why Plextor's M6S regains some of its performance, while the M6M does not.
Latency
In this test, we take the same Adobe Photoshop (Heavy) trace and use average read and write latency to compare a handful of other notable SSDs.
We shouldn't see read latency suffer due to write degradation. And that's partially why drives like Intel's DC S3500 yield flat lines; no amount of writing upsets it.
Plextor's M6es do improve as the Recovery Phase begins. I actually think that the degraded latency measurements have to do with the controller working through garbage collection and data rotation schemes, rather than a direct result of heavy write activity.
Just look at those SanDisk and Intel drives, though. This story gets even better in the next chart...
Intel's SSD DC S3500 and SanDisk's X210 are simply exquisite. Meanwhile, Plextor's M6M doesn't get a lift in average write latency, even when it's given a chance to recover. The M6S does, however, clean up well.
Despite the impressive bandwidth shown in the previous chart, the M5 Pro demonstrates interesting latency. At the end of the day, both it and the M6S finish in roughly the same place.
The Bottom Line
As we know, there is no one number able to convey the totality of an SSD's performance. PCMark 8's Storage Consistency test simply reinforces that fact. Drives don't deserve to be avoided for struggling in this benchmark, nor should any SSD be given an award for doing well. Rather, this becomes yet another tool in our suite for evaluating storage, even if it's something you might not want to mess with at home.
Why not? Well, the process isn't a simple one. There are 10 traces, and one play-through (plus round-specific preconditioning) is one phase. After 18 phases, you're presented with a shocking amount of data. That makes illustrating the information fairly difficult. I believe focusing on the Adobe Photoshop (Heavy) trace gives us a happy medium, though if you'd like to see us do something different than the above three metrics, feel free to reach out in the comments section.