Hardware Confusion 2009
[Page 3] Choosing Components
I've been quite successful building PCs that have met or exceeded my expectations over the years, but I've also made my fair share of mistakes and had some bad experiences along the way. So before getting into the specifics of my new system, in this section I provide some general guidelines for how to go about selecting the right components, drawing upon the more successful methods I've used in building various systems.
Planning the Build & Prioritizing
It's very easy to get caught up in the hype and excitement of a newly released graphics card or CPU for example, and impulse buy a component. Let me tell you that this is most definitely not the way to put together a good PC. As boring as it may sound, building a good PC requires taking some time to plan and think rationally about circumstances and your options. Even if the component you're buying is superb on its own, it can be a complete waste of time and money to rush out and buy the latest high-end graphics card for example if your PSU can't power it properly, or if your CPU or motherboard will bottleneck its performance.
The very first step is to broadly map out your needs and priorities. This may seem like a silly exercise at first, because it should be obvious that everyone wants a fast, stable, good looking, quiet, long-lasting and relatively inexpensive PC. However the reality is that when you start looking around at various components, you'll immediately find that you have to compromise and make tradeoffs. To simplify things, I've broken the decision down into six distinct priorities: Price, Performance, Stability, Longevity, Noise, and Aesthetics. You don't have to write down an actual list of needs and priorities, but spend some time thinking in detail about the rough order in which you wish to place these six priorities for your next upgrade. I'll illustrate this by listing my own needs and priorities for my new build. My needs are relatively straightforward: I need a PC which I can use for high-end gaming, for writing, and for general Internet usage. Seems simple enough, however the devil is in the detail. I put together the following set of priorities in my head, which I attempted to stick to throughout the purchasing process, presented in order of importance:
1. Stability: This is my absolute number one priority in any build. I firmly believe it's totally useless to have a fast PC if it crashes regularly or constantly has strange problems or glitches. However different people have different definitions of 'stable'. Some people believe that only getting a Blue Screen of Death (BSOD) once a week is stable; some believe that playing a game for a few hours before it crashes is stable; some believe that reformatting and reinstalling Windows every few months is normal for a stable system. I disagree. Stable is stable, it means you switch on your machine, it behaves in a completely normal and predictable manner until you shut it down, every single day, day after day, in hot or cold weather, regardless of whether you spend hours gaming or just browsing the Internet, for years at a time if need be.
2. Longevity: This is a high priority for me, because I like my systems to last. The primary reason is that a long-lasting system ultimately results in better value for money, so in a way it's related to the Price priority covered further below. But it goes beyond monetary considerations - buying future-proof gear and holding on to it longer not only lets you enjoy excellent performance both now and into the future, it also lets you totally bed down and familiarize yourself with every aspect of your system. In the first few months you'll inevitably be sorting out minor issues of various kinds, as is common with any new build. But there's nothing worse than getting everything just the way you want it, getting to know your system inside and out, optimizing it completely - and then having to go through the entire upgrade process again in 6 - 12 months' time because your system collapses in some way or is unable to provide sufficient performance. Ideally I'm aiming for a system which lasts at least 2 years without significant modification or upgrades, longer if possible.
3. Noise: This time around, rather than being further down my list, I made having a quiet PC a higher priority than performance. Even though my last PC really wasn't loud as such, it was still louder than I would have liked, particularly during summer. I felt I could do better with my latest attempt. An appropriate level of noise is subjective; one person may consider the equivalent of a vacuum cleaner sound coming from their PC to be fine, while another might consider even low-level fan noise to be too much, and want absolute and total quiet. I don't want to go to extremes, so my aim is to get an overall level of system noise similar to that which you'd hear from the stock CPU cooling fan at idle. Definitely no hair dryer noises or sounds like that of a jet taking off.
4. Performance: Performance is once again a subjective term, so in my case I should clarify that firstly I want a system which is balanced and completely responsive, whether during desktop usage or in games. I want to ensure smooth FPS at maximum or close to maximum settings in every game at my monitor's native resolution because I like my eye candy and I don't want to sacrifice it. However by virtue of placing performance more than halfway down my list of priorities, I won't tolerate potential instability or lots of noise in return for extra performance.
5. Price: This may be the number one consideration for some people, but in my case, given the computer is such a central part of what I do and what I'm interested in, it came towards the bottom of the list. That's not to say I'm rolling in money, because the opposite is actually true right now. The reality is however that if I want a machine which meets all of my other priorities, I have to spend a fair bit of money. Anyone who thinks they can buy a cheap computer which is fast, stable and reasonably future-proof is dreaming, it's just false economy. Skimping on a build will usually result in plenty of headaches and buyer's remorse. On the other hand, the fact that price is still in my list of priorities means that I'm not suggesting that money is no object. In fact this time I've made a point of not spending extra on luxuries and features which only impact marginally on performance at best.
6. Aesthetics: I consider it reasonably important that my entire system look coordinated. I don't like looking at a patchwork PC. In particular I want something that looks mature and tasteful, not childish or over the top. Ultimately though I won't sacrifice stability, longevity, quietness or performance, nor will I pay through the nose in my pursuit to achieve the right look, which is why aesthetics is last on the list of priorities.
Remember that this list of priorities isn't an iron-clad set of guidelines. It just allows you to put some thought into what sort of things you're willing to sacrifice, and reminds you of what's important when choosing components. How well I achieved these priorities in practice, and the tradeoffs I had to make, will be covered in more detail as I discuss my specific component choices in the My Choices section later on.
Deciding What to Buy
Once you have a rough mental plan of the priorities for your system, it's time for the hard part: choosing the actual components to buy. There is no easy method of doing this, it requires a straightforward but fairly tedious technique commonly known as Research. To put things in perspective, my latest system took me over two weeks of research before I started buying any components, and several more days of incidental research to iron out minor issues and ensure everything was working optimally. A conservative estimate of how much time I put into researching from start to finish would be around 30+ hours. Hopefully this article can help you cut down on your research time, but in the end there is no substitute for reading widely and thinking at length before choosing your components. Some suggested research avenues are provided below:
How and where you research your components depends on your general level of PC knowledge, and how well you keep on top of the latest changes in the PC world. For those who are less experienced and/or do not keep thoroughly up to date with the latest PC hardware changes, I recommend checking various Wikipedia articles if you're having a hard time understanding new technologies. These will give you a basic grounding in these topics, as Wikipedia is generally quite accurate on technical topics. You should then focus your initial purchasing decisions on the CPU and GPU (graphics card) in particular, as these two components have the biggest impact on performance and longevity. Start off by looking through the following four articles on Wikipedia:
The articles focus on Intel and AMD CPUs, and Nvidia and ATI GPUs, since Intel, AMD, Nvidia and ATI are the four most popular manufacturers of these components. The comparison tables in these articles may seem confusing at first glance, but even if you're not able to understand how the different products stack up against each other in terms of technical specifications, since the tables are arranged in chronological order from oldest to newest you should be able to quickly find the product names for the latest generation of CPUs and GPUs, and most will have links you can follow to specific Wikipedia articles which provide more details on them. More importantly, the list allows you to see if any newer generation of GPU or CPU is 'around the corner' so to speak, either because it will be named in the tables with a tentative release date, or discussed in the relevant article for the current generation.
For example, if you're curious as to what the latest graphics card is from Nvidia, but also want to know when the next generation of GPU will be released, first go to the Nvidia GPU comparison article, then under the Desktop GPU section find the newest GPU date listed, which in this case is the GTX 295 and GTX 285 cards, dating 8th and 15th of January 2009 respectively. Since there's no direct link to a Wikipedia article for each of these cards, instead we can use the GeForce 200 series link at the top of this particular comparison table, which leads us to an overview article detailing the latest GeForce 200 series GPUs. Within this article there is a sub-section called 'Future', which provides two very useful leads: "The D12U is expected to be released in late 2009." and:
Nvidia will be launching GT300 cards as its flagship model. The GT300 launch is projected in Q3 or 4 of 2009. It is reportedly DirectX 11 compatible. A 40 nanometer manufacturing process will be utilized, resulting in higher efficiency across the board. It is also rumoured to feature GDDR5 RAM.
Now all that remains is to search Google for the terms D12U and GT300, and variants thereof, such as Nvidia D12, Nvidia GT300, GeForce GT300, and so forth. Very quickly you will find articles which speculate on what these upcoming GPUs may offer, and when they might be released. Do all of the above for ATI cards as well, and you will have not only a good source of details for what the current generation of mainstream GPUs offer, but also some information on the next generation, which is an important factor in considering whether to delay your upgrade or not.
Regardless of how you find out what the latest generation of CPU and GPU products are, the next step involves investing quite a lot of time in searching around on Google to find as many legitimate reviews as you can for these products. For example, Core i7 is the latest generation of Intel CPU, and Phenom II is the latest generation of CPU from AMD. Using either of these terms on Google will yield page after page of search results. Your mission, should you choose to accept it, is to then browse through these results and find all the genuine review articles you can, and read as many as you possibly can before your eyes explode. It may not be a lot of fun, but when it comes to reviews, the more you read the better. The reason for this is that quite frankly I haven't found a single review site which has provided what I would consider to be an all-encompassing and 100% accurate review of any product. Most review sites receive free samples of the products they're reviewing, hence they have to maintain good relationships with the manufacturers if they want to keep receiving samples and get direct tech support. This means that quite often they will completely ignore or gloss over minor (and sometimes major) faults or issues. I discuss a very clear example of this phenomenon under the Graphics Card component in the My Choices section later in this article.
No matter how much you trust a particular tech site, you will not get the full picture until you read at least two or three fairly substantial reviews of each particular product, more if you can find them. You'll soon see that certain inconsistencies and sometimes contradictory statements will pop out at you. Reviewers' subjectivity and bias aside, the test system setups, environmental conditions, and benchmarking procedures used will all vary. Some reviews can also be downright inaccurate or contain simple mistakes which skew the results. The only way to prevent yourself from being mislead is through reading many reviews and weighing up the results in your own mind, coming to a conclusion you feel is reasonable based on a balance of probabilities.
One of the major reasons why the CPU and GPU are a good starting point is because once you have a firm idea of which CPU you might like to aim for in particular, and the GPU to a lesser extent, the rest of the system starts to build itself so to speak. For example if you want a Core i7 CPU, at the moment you can only buy an X58 chipset motherboard and DDR3 RAM, not to mention a Socket LGA1366-compatible CPU cooler if you want aftermarket cooling. This automatically narrows down your choices. If you also want to opt for SLI or Crossfire, or triple SLI for example, this will again narrow down your choices of motherboard and cooling solutions even further. If you want a particular GPU then that will determine what the maximum resolution, and hence maximum screen size, of your monitor should be, and so on. So focus on the CPU and GPU to start with, and then revise your decisions if need be after compiling a basic list of associated components and examining the initial estimates for the cost of the whole package.
I have to stress, there just isn't any viable substitute for researching and thus getting a good understanding of the components you wish to buy. This will not only help you buy the right components for you, but it also allows you to understand how to get the most out of them and also be across any quirks or minor issues they might have, and any precautions you might need to take before or after installation for example.
While searching Google for professional reviews of various products you will undoubtedly see a wide range of user-based reviews and discussions on various forums and product feedback sites. This information is a useful supplement to professional reviews and Wikipedia articles, but it is most certainly not a replacement for them. The reason for this is that user reviews and advice have a potentially high degree of bias, and also suffer from varying levels of inaccuracy. Don't fall into the trap of just asking others what to buy out of laziness.
Most people who post tech advice on forums do so to help others, however there are also other underlying motivations which can confuse things. People love to be seen as being tech-savvy or an 'expert', so quite often they will regurgitate (mis)information they have heard from someone else or read somewhere, without any idea of its validity, in an attempt to appear knowledgeable. In some cases they might even make up or guess at something, but put it across as a statement of fact, again so as to appear knowledgeable and thus earn peer respect. Worse still, some people have the urge to deliberately distort or omit facts when giving advice or feedback in an effort to reinforce the apparent wisdom of their own purchasing decisions and brand biases.
A classic demonstration of these behaviors can be seen in any user discussions of graphics cards and graphics drivers. It may seem amusing to non-techies, but it's (sadly) true to say that some computer enthusiasts can be blindly loyal to a particular brand of graphics card, whether ATI or Nvidia, and will routinely distort the truth or speculate without any basis in fact to ensure that their choice of brand appears to be the best. This 'fanboyism' makes it very difficult to sort out fact from fiction when graphics card-related discussions take place. Ask any two 'experts' which graphics card to buy and you might get two totally different and conflicting answers.
Furthermore the level of technical knowledge (or lack thereof) among many forum users and blog writers means that all sorts of unsubstantiated statements and erroneous attribution of various problems occurs. One person will blame a particular graphics card and its driver for a game crashing, while conversely someone else with similar hardware and drivers will blame the game. Some people may label a particular component as being unstable or of poor quality. In all cases, it could well be that their own ignorance or negligence has resulted in destabilizing or harming the system (e.g. through overclocking, overheating, incorrect software or hardware configuration, etc.).
In short while user-based reviews and discussions have the potential to be very valuable in finding out whether there may be any issues with particular components, the problem is that they can also be highly inaccurate or misleading and result in even greater confusion, so don't rely on them as your primary source of research, and most definitely don't ever base a purchasing decision on the word of one or two people (and that includes me), however much of an expert you believe them to be. Do your own research thoroughly and then use the advice and feedback of others as an additional source of information, allowing you to fill in the gaps and get a better feel for various issues which professional reviews may not touch upon.
Some of the components you may wish to purchase will be available in various brands, the most common being graphics cards. There are two major graphics card companies which make most of the standalone graphics cards on the market: Nvidia and ATI. However there are many different manufacturers - called OEMs - which use Nvidia and ATI-designed chipsets to produce their own brands of such cards. For example, a graphics card using an Nvidia GeForce GTX 295 chipset may have as many as a dozen different possible brands under which it will be sold. Companies like EVGA, ASUS, XFX, Gainward, Gigabyte, Palit, Leadtek, Zotac, Inno3D and Galaxy will offer what appears to be essentially the exact same card for sale, often at significantly varying prices. Thus people might wonder not only which model of graphics card to buy, but which brand of that particular model.
In the case of graphics cards, this is actually an easier choice than would first appear, because at the moment most brands of graphics cards based on a particular chipset will usually be virtually identical in terms of general build quality and performance. This is because the chipset makers Nvidia and ATI have what is called a Reference Design, a blueprint which they provide to OEMs, and in some cases, even supply the specific materials required to build these cards, so as to maintain a minimum level of quality. Note that some factory overclocked cards may use better quality components as well as providing better performance, but this varies on a case-by-case basis, so check reviews.
There is however one potentially significant difference between brands: warranty. Different brands will have different lengths of coverage and various terms and conditions for their warranties, and this is probably the only real reason to choose one brand over another. Importantly, you should research to see if there is any user feedback regarding warranty claims with the brand you've chosen, because sometimes seemingly generous warranty provisions are countered in large part by lengthy delays in getting your card repaired or replaced under warranty. Also keep in mind that physically modifying or overclocking most products will void their warranty.
Something else worth keeping in mind is that there is no single brand which is guaranteed to be problem-free. Blindly throwing lots of money away buying expensive brands just to make sure nothing goes wrong doesn't work. It depends on the particular component as to which brands may be better at any particular time. For example I consider the ASUS brand to be a reasonably sound assurance of quality and good features when it comes to motherboards, based on my experiences with previous ASUS boards I've owned. However even ASUS motherboards can and do cause problems for some people. That's why the warranty terms, and the way particular manufacturers handle warranty returns are again probably the single most important determinant of which brand is "best". You can run into problems with virtually any brand; it's how a particular manufacturer honors their warranty, and how much time, money and effort it takes you to get a repair or replacement that makes the real difference. I provide a concrete example of this under the Monitor component in the My Choices section.
Anyway these are just some general guidelines and considerations I try to adhere to when choosing components. In the next section I start my lengthy blow-by-blow account of the specific components I chose, and the reasons why.