re: Build the Best Gaming PC Your Money Can Buy: A Detailed Guide (Updated Sep 2014)

------------------------
++++ALERT: You are reading an out-of-date version of the guide and wasting your time. Read the PDF for the most accurate up-to-date info.It's best to download the PDF and use a proper PDF reader. Google's formatting of PDFs breaks all of the links. Link to directly download the PDF. I have stopped updating the text in the thread because the forum's limited code makes it far too time-consuming to change images and add text.++++
------------------------

The CPU

In the graphics card arena, the AMD vs. NVIDIA war can get ugly. But it still doesn’t hold a candle to the vitriol exchanged in CPU debates—AMD vs. Intel. Those are the two CPU companies under consideration here, and I won’t attempt to convince you that one is better than the other in all situations. My goal will be simply to educate you on why it’s not that big of a deal in a gaming PC at the budgets we’ve set here. For that reason, I won’t spend much time discussing all of the different CPU options out there.

First, let’s clear the air regarding any preconceived notions about clock speed. It’s a common argument used among uninformed individuals who claim that AMD’s $200 4GHz CPU outperforms Intel’s $320+ 3.5GHz CPU. False.

When you consider everything that makes up a “CPU” (cores, cache, pathways, registers, etc. etc.), there's no point in comparing clock speeds. Even fairly educated enthusiasts, me included, find it difficult to make a purchase decision based solely on a CPU’s technical specifications. It’s far more practical to compare CPUs by looking at quantifiable real-world performance, benchmarking, and pricing.

====///====Working Harder vs. Working Smarter====\\\====
When a processor does its job (the workload), it breaks the load into stages/phases, then further divides those stages into units that can be processed more than one at a time. Software might provide commands/instructions in a linear manner, but a CPU can prioritize those instructions differently. Let's say a program sends out instructions X, Y, and Z. The CPU can divide those instructions into subunits (say, X1, X2, X3, Y1, Y2, and so on) and execute them in the most efficient order (which might be X2, Z1, X3, X1, Y2, etc.). That’s an oversimplified example, of course. It’s much more complex in reality.

The design of the CPU makes the clock speeds unrealistically comparable between different CPUs. It depends on how much can be done in what order, the divisions and subdivisions of labor it can provide, and the speed at which each of those units/subunits can be executed (some of it has a lot to do with the register’s capability, which stores commands, and the cache, which stores data associated with those commands).

Perhaps more difficult to explain is the difference in designs that give one CPU the ability to break down commands into more and therefore smaller subunits than another CPU because it extends beyond the number of cores. However, the end result is that the clock speed dictates how quickly those subunits will be executed, and the architecture/design dictates how many subunits can be executed per cycle—or, instructions per cycle (which is a blurrier distinction because I'm not an engineer). More subunits per instruction means smaller workload per unit, which means faster processing of that subunit at the rated clock speed. On top of that, less workload per subunit also means more subunits per clock cycle.

At this point, Intel takes the lead in doing more per clock cycle, and the differences can boil down to material choices, die size, transistor count, and overall design. But it can sometimes require a trade-off. More work at a slower pace, or less work at a faster pace? It largely depends on how efficiently the software can use the architecture, which is why CPU specs must be taken with a grain of salt and it is necessary to examine relevant benchmarks. And what benchmarks are relevant to us? The gaming ones, of course. More on that later.

====How Many Cores?====
So, what about core count? That 4GHz AMD has 8 physical cores, and the Intel 3770K/4770K only has 4 physical cores.

More cores are certainly better, but there is sometimes a misconception about AMD’s core count. The current 8-core AMD FX series are not exactly “true” 8-core processors. Their architecture uses what’s called a modular design, with 4 modules containing 2 cores each. Each of those 4 modules contains only a single floating point unit (FPU), which is responsible for carrying out computations. The cores on each module must share the FPU, as well as other components with the module, such as its cache. Granted, this does not mean it’s really a 4-core CPU. It still has 8 physical cores; they are just somewhat crippled by the need to share certain components.

A 4-core Intel CPU with hyperthreading has 4 physical cores and 4 virtual cores. As I said before, Intel’s current CPUs are able to do more work per clock cycle per individual physical core. When you add hyperthreading to the mix, you completely optimize the work output from those physical cores. Hyperthreading actively searches for any core that is currently in an idle state waiting its turn, and uses that idle core to immediately process the next set of instructions. It ensures that no cores are needlessly idle in applications that are actually coded to acknowledge multiple cores, or in general multitasking.

If we were building a 3D modeling, video editing/encoding machine, a quad-core hyperthreaded Intel i7 would perform better than AMD’s 8-core FX CPU and use less power—but it would also cost over $100 more than the AMD. However, since we’re building a gaming PC, all we really need is a CPU that won’t be the performance bottleneck in the games you enjoy.

====///====What Makes a CPU Bottleneck Game Performance?====\\\====
A CPU bottleneck means the processor is holding back the graphics card from rendering more frames per second. Meaning, the graphics card is not living up to its potential, and therefore neither is your gaming experience.

This bottleneck can occur because not only is your graphics card working hard to render what you see on the screen, your CPU is handling what’s actually happening in the game. CPUs are particularly good at handling physics, so that would include character movements, explosions, crashes, shooting, etc. in a dynamic environment.
Some games are CPU-intensive simply because they were coded that way. Skyrim, for example, relies heavily on your CPU because certain rendering components (such as shadows) were coded for the CPU.