Solid state drives (SSD) appear to be as unreliable as traditional hard disks. In fact, they're marginally less reliable: Taken as an average across models, 2.05 percent of SSDs got returned as non-functioning, compared to 1.94 percent of hard disks.
That's according to French hardware review and optimization site Hardware.fr, which published its sixth-monthly report into hardware reliability.
Given free access to an unnamed online retailer's sales and returns database, it was able to measure the failure rates of hardware sold between October 2009 and April 2010. The data was sampled in October just passed, giving a maximum potential usage period of 12 months.
This time around, hard disks appear to have slipped a little in reliability compared to the previous survey six months ago, when the failure rate was 1.63 percent. In other words, the difference might have been even greater. (This is the first month for which Hardware.fr has statistics on SSDs.)
Solid state disks rely on flash memory chips and contain no moving or magnetic parts, unlike with traditional hard disks. Therefore it's always been assumed they should be much more reliable in the short to mid-term, although all flash memory has a finite life span that's shorter than that of hard disk counterparts.
Of the five traditional hard disks manufacturers included in the survey--Maxtor, Western Digital, Seagate, Samsung, and Hitachi--only one had a higher failure rate than that of the worst-performing solid state disk brand--SSDs from OCZ saw 2.93 percent failure returns, against Hitachi's 3.39 percent.
Intel's SSDs came out best, with just 0.59 percent failure rates.
When it comes to RAM, it perhaps comes as no surprise that high-performance modules see the highest returns. These rely on state-of-the-art components that are then pushed to the very limit of what they're capable of. However, the failure rates for the worst offenders are massive, with more than one in 10 modules proving faulty.
In the top spots for individual memory product returns are a staggering 15.08 percent failure rate for the Corsair, and a better but still worrying 11.28 percent rate for the OCZ module.
Taken as brand averages, Corsair fares better. OCZ is by far the worst with 6.76 percent failure rate, followed by second-place G.Skill, with 2.73 percent. Corsair comes out third taken as a whole, with just 1.41 percent of its modules failing.
Elsewhere there are fewer surprises. Motherboards and graphics cards from virtually every manufacturer seem to average a failure rate of up to 3 percent, while power supplies offer a little more variation, with rates up to 3.3 percent. If you're looking for an ultrareliable power supply purchase, the Thermaltake EVO Blue 550W had zero returns, according to the survey.
And the most reliable component in the PC is the CPU, with a failure rate of 0.18 percent. This won't surprise anybody with experience: I've never had a CPU fail on me, at least beyond my own inability to provide adequate cooling. Indeed I have computers of between 10 to 20 years old whose processors work fine.
Ultimately, however, this survey doesn't provide high quality data. The biggest issue is that the results refer to products sold between October 2009 and April 2010. Some might possibly still be on sale now, but many will have been superseded by newer ranges. This has a particular impact when considering the memory modules figures, for example; the list of worst-performing memory modules includes several DDR2 performance modules. A contemporary list would no doubt involve almost exclusively DDR3 technology.
All technologies need time to bed-in in order to get optimal reliability. Similarly, SSD is a young technology and when the survey data was taken, it was even younger. It will be fascinating to see the results of the next survey in six months time to see if the trend continues.
Keir Thomas has been writing about computing since the last century, and more recently has written several best-selling books. You can learn more about him at http://keirthomas.com and his Twitter feed is @keirthomas.