THG's Secret Test Sauces Revealed

Preparation

Most of the programs on our list are commercial software packages that you have to pay for. You thus might not be inclined to spend the money to buy all of the benchmark programs we detail here. Nevertheless, your choice of benchmark software should always be based on which components you want to test since every program that is suitable for benchmarking can be assigned to a particular category.

Different benchmarks are more appropriate, of course, for certain components. For example games, especially 3-D titles, are excellently suited to test the performance capacity of your CPU and graphic card.

In most cases, it doesn't even matter what kind of hard drive you use. However, audio or video encoding programs (Auto Gordian Knot, Lame, MP3 Maker) always need a lot of computing power, and in some cases a fast main memory also makes a difference. Occasionally the hard drive capacity also influences results; for example, when the data stream being written has reached a certain volume, such as with MainConcept MPEG Encoder.

Likewise, software such as WinRAR for compressing files requires high computing power and decent storage capacity. While the computer is working on the actual compression, the hard drive capacity doesn't matter. Professional graphic applications such as 3DS Max, Cinema 4D or Lightwave require a fast CPU more than anything else.

Moreover, the settings you've chosen influence the results and can sometimes act as a bottleneck. If, for example, you want to compare two motherboards or processors, nearly all 3-D shooters lend themselves to this task. However, if you've set the screen resolution too high, the graphics card will slow things down and invalidate the results. This effect can be spotted by the fact that products in the same category that would be expected to perform quite differently (for example, motherboards) then produce similar results. Though, admittedly, it's not unusual these days for most core logic products (graphic chips, chipsets and so on) to produce similar results, so this can't be taken as a hard and fast rule.

Particularly when you're comparing two interchangeable components, there are two basic principles to which you need to adhere. The first is to never compare apples to oranges. The basic components used in the test should never be changed in the middle of the test. If, for example, you're comparing several types of graphics cards, then the motherboard, CPU, RAM, hard drive, etc., shouldn't vary. Likewise, it's taboo to change the settings in the middle of the test. Second, you should try to have comparative values for similar, well-known products at hand. The reliability and overall value of your benchmarks have a lot to do with how much you follow these principles.

An entirely different concept of benchmarking involves comparing two completely different computers. In this case, it is okay to compare apples to oranges.