We all know that the processor, or Central Processing Unit (CPU) as it's more formally known, is the brains of the computer - the part that actually does the calculations. However, when you look at a specification for a computer you're thinking of buying, you'll probably see something like the examples below, and it's not as obvious what all of the parts mean and how you compare one with another.
Intel Atom Processor (1.6GHz, 512KB L2 Cache, 533MHz FSB)
Intel Celeron 550 Processor (2.0 GHz, 1MB Cache, 533 MHz FSB)
Intel Core 2 Duo Processor T5750 (2.0 GHz, 2 MB L2 Cache, 667 MHz FSB)
Intel Core 2 Quad-Core Q9450 Processor (2.66GHz, 12MB cache, 1333MHz FSB)
The format of each of these examples is: Manufacturer, CPU type and number of cores, (clock speed, cache size, bus speed). While a specification may be worded in a different way, it usually contains the same basic pieces of information. These pieces are explained in the following sections.
In the examples above this is Intel. The other big manufacturer is AMD. Both companies have strong advocates, but in reality determining which is best is like trying to hit a moving target - one may be better than the other at a certain point in time, but then the other will release a new set of chips which puts it temporarily ahead, and so on.
Performance at the very top end isn't likely to be a factor for anybody reading this anyway. In the range that we're concerned with (good processors for everyday usage), both provide good products which will meet your needs. So, our advice would be to ignore whether the chip comes from Intel or AMD, and make your decision based on other factors.
CPU Type and Number of Cores
In the examples above, this is Atom, Celeron 550, Core 2 Duo T5750 or Core 2 Quad-Core Q9450. This tells you three things:
The product line of the chip (Atom, Celeron etc.), which is much like a particular model of car. In the same way that Ford produce numerous versions of the Focus, all of which share the same basic design but have different options and levels of performance, the chips in a product line such as Celeron will all share the same architecture but will vary in terms of speed and other characteristics.
The number of cores. Historically, a single CPU was packaged into a single chip, and if you wanted more computing power you either used a faster CPU or added an additional chip. Now, it's increasingly common for a chip to contain two or more CPUs, each of which is capable of processing data independently. In the examples above, the Atom and Celeron chips contains a single CPU, the Core 2 Duo contains 2, and the Core 2 Quad-Core contains four separate CPUs. Similarly, Athlon X2 chips contains 2 cores, X3 chips contain 3 and X4 contain 4.
You might think that this means that a 2-core chip will perform twice as fast as a single-core chip of the same speed, but unfortunately it's not that straightforward. If you're running multiple applications simultaneously, or if your computer is performing a task which can be cleanly separated into parallel tasks (which depends on whether the software has been designed to do this), then the performance will be significantly better than an equivalent single-core processor. Otherwise, if you're running a single application which has not been designed to take advantage of multiple cores, then you'll see little or no improvement over an equivalent single-core processor.
The model number (T5750, Q9450 etc.), which is just a unique identifier assigned by the manufacturer to that particular version of the chip.
This is the rate at which a processor performs instructions. Obviously the faster this happens the better, but it's not the only factor that affects overall performance, and comparing processors purely on the basis of clock speed can be misleading. The architecture of the chip, how many separate instructions it needs to perform to implement complex tasks, the size and types of memory cache, the amount of RAM, and many other factors also have a significant influence.
In the late 1990s and early 2000s, Apple used the PowerPC line of chips in their machines. While these had slower clock speeds than the Pentium chips being produced by Intel, Apple claimed that they actually performed better in some real-world situations, and used the term "Megahertz Myth" to describe the misleading impression that speed was everything. Apple presented data showing that in a set of tests an 867 MHz PowerPC G4 chip performed up to 80% better than a 1.7 GHz Pentium 4 (see The Megahertz Myth).
For multiple-core chips, the clock speed quoted applies to each individual core. So, for example, and Intel Core 2 Duo Processor T5750 (2.0 GHz, 2 MB L2 Cache, 667 MHz FSB) has 2 cores each running at a clock speed of 2 GHz.
A cache is a place to temporarily store data that needs to be accessed frequently. Caches can be accessed very quickly, so this is much more efficient than having to repeatedly retrieve the data from disk.
There are multiple levels of cache in a typical computer. Generally the size of the cache increases as you go down through the levels, while speed of access decreases (so the first level is the smallest but has the fastest access, the next is larger but not quite so fast, and so on). Typical cache levels are:
The original definition was that this was memory that was on the processor chip. With the advent of multiple-core CPUs, this is now not just on the chip, but part of the individual CPUs. It has the fastest access speed (because it's closest to where the processing is done), but is generally very limited in size.
The original definition was that this was memory that was outside of the processor chip but still part of the motherboard. With the advent of multiple-core CPUs, this is now part of the chip, but is separate from the individual CPUs and is shared between them. It is typically much larger than L1 cache, and can be accessed quickly, but not as fast as L1 cache.
Now that L2 cache is typically part of the chip itself, it is possible to have a third-level cache which is separate from the chip but on the motherboard. It's typically larger and slower than L2 cache, but faster than RAM.
RAM (Random Access Memory):
While it has a different name, this is another level of cache. It's contained in separate memory modules which can be inserted into your computer, and is typically much larger than the L1-L3 caches (up to 4 GB is now common) but cannot be accessed as quickly (although it's still much faster than accessing data from a disk).
Caches are checked sequentially - the L1 cache is checked first, if the data is not present then the L2 cache is checked, then the L3, then the RAM. Since access speed decreases at each level, bigger caches obviously help to improve overall performance.
Front Side Bus (FSB) Speed
The last part of the specification in the examples at the start of this page referred to things like "667 MHz FSB". FSB stands for Front Side Bus and is an electrical pathway which connects the processor to other components like the RAM and the hard disk. Obviously if the rate at which data can be transferred over the FSB is not fast enough then it becomes a bottleneck, and it doesn't matter how fast your processor is, because you can't get information to and from it quickly enough.
Taking into account all of the factors which influence overall performance, it's very difficult to compare one processor with another and determine which is best. It depends on so many things, and also varies depending on the specific task that you're trying to perform.
If you are interested in objective comparisons, then there are sites that show the results of various benchmarking tests. Our favorite is Tom's Hardware which, under the Charts -> Processors section, shows the results of various benchmarking tests for a wide range of processors.