Table of Contents
At Puget Systems, we record a huge amount of data for each and every system we sell including benchmarks, BIOS screenshots, thermal images, and system photos. In fact, much of this data is published on our website and can be accessed through our part information pages. Simply view a part information page, scroll down to the Links section, and you can view BIOS Screenshots, Thermal Images and Product Photography from recent systems we built.
While this data is nice, one of the most important things we track is the failure rates of individual components. Reliability is one of our primary values, so this data is invaluable for tracking both individual component, product line, and overall brand failure rates. With 2013 coming to a close, we thought we would run some reports and share what hardware we found to be the most reliable in 2013.
Since we are a custom computer company and do at times special order in components to meet a customer's specific needs, there is one stipulation we are imposing on what hardware we will be allowing into this list. Specifically, we are only considering high volume items from our product line so that we have a large enough sample size to make an informed call on the reliability of the component.
With that said, let's take a look at one of the most at risk components in a computer: the motherboard.
A reliable motherboard is essential in a high quality computer. Not only is a motherboard very difficult to swap out, but the effects of a poor quality motherboard can be far reaching and difficult to troubleshoot. This is complicated by the fact that motherboards are one of the most complex components in a computer. There are SATA, USB, fan, and network controllers as well as physical ports, audio chips, and everything else that is needed to inter-connect every component in your system. This is a huge number of small parts that have to work perfectly together, and any one of these could potentially have a problem. If there is a single dead USB port, slight static over the audio, or the voltage levels are measured outside of norm, it does not meet our standards and is considered to have failed.
Because of this, motherboards have one of the highest overall highest failure rate of any core component with 1 out of every 20 motherboards failing for one reason or another. This may seem like a high failure rate, but the silver lining is that nearly all of these failures we catch in-house before the system is shipped to the customer. In fact, motherboards as a whole only have a .54% (or about one out of every 200) when you only look at post-shipping failures.
At Puget Systems, every motherboard must pass an extensive qualification process but there is no substitute for reviewing hard evidence after offering the product for a period of time. We have to simply keep an eye on our failure reports and quickly move on any trends we may see. From these failure reports, we found four motherboards that had absolutely zero failures in 2013.
Interestingly, the only two Intel-based motherboards are from Asus' Workstation class of motherboards and both are the micro-ATX version. Surprisingly, we had no Intel-based desktop boards that had a low enough failure rate to warrant inclusion in this list. Instead, this is the first year that two AMD-based motherboards have made our most reliable hardware list, both using the AMD FM2 platform.
Impressively, even though more and more technology is moving from motherboards to CPUs (memory controller, voltage regulator, etc.), the overall failure rate for both Intel and AMD CPUs dropped dramatically in 2013. Whereas 2012 had an already impressive overall failure rate of .47%, 2013 was even better at just .39%. So instead of listing all the different CPU models that are extremely reliable, we are simply going to say that every CPU made in 2013 is incredibly reliable.
At Puget Systems, we almost exclusively use Kingston RAM in our computers. We do so because we know from experience that they are extremely reliable. Any time we tried any of the other prominent brands (when Kingston was either in shortage or did not offer exactly what we wanted), we almost always ended up moving back to Kingston once we were able. This is primarily due to the fact that as a whole, Kingston is as much as three to four times more reliable than other brands.
Standard desktop RAM as a whole has an low failure rate (.84%), which is a bit lower than last year's ~1% overall failure rate. Also better than last year is that we had one model (Kingston HyperX DDR3-1600 4GB Low Voltage) that has a 0% failure rate. Interestingly, the other model that had an extremely low failure rate was also a 4GB model, only from Kinston's more mainstream ValueRAM line.
ECC RAM is specifically designed to have a low failure rate, but it is a bit of a surprise that we had absolutely no ECC RAM failures in 2013. Even the huge 32GB sticks had not a single problem. This means that for 2013, ECC RAM has a completely perfect record
For laptop SODIMM RAM, there were two models that we found to be highly reliable. Interestingly, both are the new low voltage RAM from Kingston that we just added this year. This is great to see since we had hoped this RAM would be more reliable since it uses less voltage than standard laptop RAM. The 8GB version does have a fairly high failure rate at almost .5%, but we decided to include it since this .5% actually only equates to a single stick having failed.
Just like RAM, we typically only use a few brands of SSDs and hard drives that we historically know to be extremely reliable. 2013 was a bit different as we switched from Intel to Samsung drives for most of our SSDs. Thankfully, this turned out to be a great move for reliability. In fact, only a single Samsung model we sold in the last year had any failures (2 total, or .7% to be exact). The other three models all had a 0% failure rate
What is great to see this year is that even the large 1TB model had no failures. Historically the larger the SSD, the higher the risk of failure. So kudos to Samsung for making some very reliable SSDs.
Traditional platter drives are at a higher risk of failure than SSDs due to the moving parts, but in 2013 we had so many reliable platter drives that we are instead going to sort by product line rather than individual drives. Out of the six Western Digital product lines we currently offer (Black, Green, Blue, Red, Velociraptor and RE), there are only two lines with absolutely no failures in 2013. The Western Digital Green line makes sense as they are designed for low power and thus create less heat. The Blue line was a surprise to us as they are supposedly the "budget" line, but they have shown that the word budget does not necessarily mean cheap in terms of quality.
With the wide range of video cards we offer (including a mix of brands), naming the most reliable model is a bit tough as we don't sell large quantities of any one card. However, of the cards we sold enough of to have a good feel for their reliability, there are a few that stand out with a 0% failure rate.
EVGA GeForce GTX Titan 6GB
Asus GeForce GTX Titan 6GB
Zotac GeForce GT 640 2GB Passive
Asus Radeon HD 7850 2GB DirectCU
Impressively, both the GTX Titan and GTX 780 – which are among the fastest cards made by NVIDIA – did very well in terms of reliability. Both EVGA and Asus made cards that had absolutely no problems at all in 2013. Outside of those, the Zotac 640 2GB Passive card also had no failures as well as the Asus Radeon HD 7850 2GB DirectCU.
This list may look very NVIDIA heavy, but part of it is that we simply sell more NVIDIA cards than AMD cards, so there are more NVIDIA models that we have sold enough of to be confident in our data. On the other hand, NVIDIA GeForce cards on a whole were much more reliable in 2013 than AMD Radeon cards. NVIDIA GeForce cards only had an overall failure rate of 3.3% versus AMD Radeon cards which had an overall failure rate of 10%
Not mentioned above are also NVIDIA Quadro and AMD FirePro cards. Both lines had very low failure rates (2.05% for NVIDIA Quadro and 2.17% for AMD FirePro), but we don't have enough data on any one model to say that one model in particular is more reliable than others.
Out of all the power supplies we sold an appreciable amount of in 2013, there is only one model that had no failures at all:
Funny enough, this is actually a model that has technically been end of life for some time now. We have been purchasing them in special batches directly from Antec specifically because they are so quiet, affordable and extremely reliable. We also want to give a special mention to the entire Seasonic X-### series which – although they had a 1.75% failure rate overall – only had a single failure after they left our shop.
So there you have it: the most reliable hardware for 2013. While there is too much data to make many broad generalizations, there are a few simple ones that can be made. First, RAM (even non-ECC versions) was much more reliable in 2013 than it has been in previous years. There are a number of reasons for this, but the largest is that DDR3 is currently a very mature product. RAM reliability will likely drop pretty heavily when DDR4 is introduced, but for now it is more reliable than it has been in years.
Second, we made a number of product line moves and additions in 2013 that turned out to be great for reliability. Chief among these was the move to Samsung 840 Pro SSDs which proved to be more reliable than their Intel counterparts (in our experience at least). Low voltage laptop SODIMM memory also was a great addition as it has shown to be much more reliable than the standard SODIMM RAM. Lastly, AMD CPU and motherboard reliability greatly improved in 2013, but AMD Radeon video cards as a whole dropped in reliability.
This wraps up the most reliable hardware of 2013, but if you enjoyed reading this article we invite you to check out the next section that we decided to tack on to this article. It is not about the reliability of specific hardware, but uses much of the same data and in our opinion is very interesting.
Tangent: Home Built versus Puget System Failure Chance
In addition to tracking simply how often each part we sell fails, we also keep track of whether that failure was found while the machine was initially built/installed/tested or if it failed at some point after the customer received the system. While reviewing these numbers, we realized that this is actually a great metric for one of the many value-adds a company like Puget Systems provides: a much greater reduction in hardware issues for the end user.
In short, the overall failure rate for each component category is equivalent to the risk of a component failing if you were to build the computer yourself. However, this doesn't take into account user error or mistakes and is reliant on using the exact same hardware we use which we have spent 13 years refining and tweaking to allow only the highest quality components. So if anything this number is overly optimistic. However, even as extensive as our testing is, we can't catch every possible problem and there are some issues will occur no matter how much testing we perform. The end user failure rate represents these failures and put simply is the risk our customers have of seeing a hardware failure.
|Hardware Category||Overall Failure Rate||End User Failure Rate|
|Motherboard (Desktop Intel)||4.71%||.47%|
|Motherboard (Desktop AMD)||4.23%||0%|
|Motherboard (Server Intel)||3.86%||.35%|
|CPU (Intel Desktop)||.35%||.2%|
|CPU (Intel Xeon)||0%||0%|
|CPU (AMD Desktop)||0%||0%|
|Hard Drive (SSD)||.48%||.19%|
|Hard Drive (Platter)||.57%||.28%|
|Video Card (NVIDIA GeForce)||3.29%||.8%|
|Video Card (NVIDIA Quadro)||2.05%||0%|
|Video Card (AMD Radeon)||10.09%||1.83%|
|Video Card (AMD FirePro)||2.17%||0%|
This is somewhat difficult to make sense of, so lets create two completely generic systems and think of it in terms of comparing the chance of a hardware failure happening on a system you build yourself versus one that you purchased from Puget Systems:
Remember, this is using an average failure rate from each hardware category, so it is about as generic as it can get. If you use hardware from this article that is more reliable, the difference in the two failure rates will go down. At the same time, if you use hardware that is at a higher risk (high frequency RAM for example) the failure rate difference will likely increase dramatically.
What this shows is that if you built an Intel/NVIDIA GeForce system yourself, based on past failure rates you have about a 1 in 7 chance of there being some sort of hardware problem. But if you purchase the exact same system from Puget Systems, this risk goes down to a 1 in 30 chance since we catch the majority of the hardware problems before you would even see the machine. Similarly, if you build an AMD/AMD Radeon system yourself, you have a a 1 in 5 chance of having a hardware problem versus a 1 in 27 chance if you purchase the exact same system from Puget Systems. In short, our data indicates that you are approximately 4-5 times more likely to encounter a hardware problem when building a computer yourself than when purchasing a complete computer from Puget Systems.
Keep in mind that these numbers are only true for Puget Systems computers/hardware and won't be true for every computer manufacturer out there. Here at Puget Systems we put our computers through a very rigorous testing process so we are much more likely to find hardware problems before the customer receives the machine than most other companies. So if you purchase from other companies – especially bigger companies that do little if any testing on their systems before shipping them to the customer – you may not actually be reducing your risk of encountering a hardware problem, or at the very least not reducing it by as much as if you purchased from a company like Puget Systems.