10 November, 2015
By Joe Callison
Computer reliability data is much like reliability data for automobiles or any other major consumer products. You may get a bad one buying the statistically most reliable brand or you may get a really reliable one buying the statistically least reliable brand. Still, you probably want the odds to be in your favor when considering your purchase. It also helps to know how a brand’s reliability is trending over several years, since any brand can have ups and downs over time.
One very large computer repair company has been publishing computer reliability data based on brand market share data and their own repair share data for several years. The premise is that a brand with 20% market share should account for about 20% of their repair business if all brands are equal in reliability. If the percentage of their repair business for a brand is much higher than its market share, then the brand is considered less reliable than a brand that is less percentage of their repair business than its market share. The reliability data is published by many media outlets all over the world. By comparing the reliability rankings of brands at around the same time of year over a several year period, you can get a pretty good indication of a brand’s overall quality.
From rescuecom.com Computer Reliability Reports —
(Rank Highest to Lowest Reliability):
|2012 Q2||2013 Q3||2014 Q2||2015 Q2|
Other interesting results come from Microsoft’s crash data sent from over a million PCs. Analysis of only hardware related crashes, excluding overclocked and underclocked systems, shows that laptops are more reliable than desktops even though they are typically subjected to rougher handling and higher temperatures.
The Microsoft crash data also shows that brand name computers are more reliable than systems built by everyone else. This is most likely due to the prequalification of components based on quality and testing of systems before delivery to eliminate early failures for brand names compared with others that include do it yourself builders. One computer system builder, Puget Systems, claims extensive prequalification and testing before delivery and calculates the chances of failure in a home built computer system at over 4 times their failure rate for Intel processor/NVIDIA graphics based computers and over 5 times their failure rate for AMD processor/AMD Radeon graphics based computers. In my own experience building dozens of computer systems for people, I know the importance of selecting components with the highest reliability ratings for the build and running the system for at least 8 hours or overnight to reduce the chances of failure after delivery.