In April, Intel released their next generation processor architecture known as Ivy Bridge, which also coincided with my PC upgrade schedule.
One of the more unique design choices I made was to go with a RAID0 solid state drive setup. The primary reason for this decision was that I already had two Intel X25-M (G2) drives available and therefore wanted to avoid any unnecessary expenditure, especially as at the time solid state drives were still quite expensive.
As you are probably aware, RAID0 can offer significant performance improvements (theoretically doubling the read / write speed), however when using solid state drives it does come with a key limitation. Unfortunately when using any type of RAID on Windows you cannot enable TRIM, which is the mechanism used by the operating system to ensure the drive stays in peak performance. Analysis has shown that the performance degradation is not easy to predict, in fact the drive type, operating system and usage will all have a bearing on the overall impact.
As I have now been using my RAID0 solid state drive setup for approximately three months, I thought it would be useful to re-run the storage benchmarks to see what performance I have lost (if any).
Before sharing the results, it is worth noting my system specification and usage patterns. The PC is running Windows 7 Ultimate x64 SP1 with all the latest updates and drivers and I have used 142GB of the available 297GB (47%). It is used daily, primarily for manipulating small files and playing the occasional game. All large file storage (media) is backed-up to a Synology NAS and therefore does not impact the RAID. The local disk (RAID0) Windows properties can be seen below:
Now let’s jump to the before (April) and after (July) benchmark results. As with my previous articles I will use CrystalDiskMark v3.0.1 x64 to validate the performance.
Before (April 2012):
After (July 2012):
As you can see from the results, the overall performance has maintained very well with the only noticeable change being the “Sequential Read” test which dropped from 586MB/s to 425MB/s. Although this looks like a significant drop, I’m not convinced I would be able to notice this difference in a real world scenario.
It’s also worth remembering that a single Intel X25-M (G2) drive can only achieve a maximum of 264MB/s sequential read, therefore the RAID0 setup is still significantly faster.
To conclude, I continue to be happy with the RAID 0 setup, especially when you compare the results to a single Intel X25-M (G2) drive. However I plan to continue to monitor performance as it’s possible that the lack of TRIM support could have a bigger impact as the drive becomes more heavily utilised.