When testing computer hardware, it is unfair to compare meek components to roaring and powerful ones. Therefore, before testing any hardware component, you must recognize which of the three main categories it falls into, entry- level, mid-segment-referred to as mainstream-or high-end. In the CHIP comparison tests, you'll perpetually notice a clear segregation of products into these categories.
Entry-level hardware offers. basic performance and serves for tasks that are a necessity in computing. For instance, an entry-level processor suffices only for basic applications such as Microsoft Word, Excel and other less demanding third party software. In contrast to that, editing or transcoding videos, encoding audio, and creating or rendering visual effects are not the jobs for an entry-level CPU. As far as gaming is concerned, entry-level graphics cards can handle only entry-level games and lower resolutions which do not require heavy rendering of graphics.
Heavy applications cause extreme stress to even mainstream hardware. For instance, a mainstream graphics card can play all basic games and handle high-end games on medium settings. Playing Crysis at 1440 x 900 with 4 X AA turned on will be a smooth ride. However playing games at higher only high-end graphics cards can handle.
There are certain factors that determine whether a specific hardware component will be able to handle a particular task. Also, to make the correct buying choice, it's important to know how components measure up against each other and that's exactly where benchmarks come into the picture.
Benchmarking includes a series of tests that stresses a hardware component to get the maximum throughput from it. In a comparison test, there are variants of the same component. Once all the variants from one specific category-say entry-level-are tested, their scores are logged. The highest score amongst the components from this category becomes the 'benchmark'.
The performance scores of all other contenders belonging to that category are compared to the benchmark score. Although there are benchmarks that are designed to check quality, quantity, physical features and so on, in this story we shall concentrate only on performance benchmarks.
Also, performance benchmarking is not restricted to computer hardware, it's a process which is also used for measuring the performance of portable gadgets such as laptops, netbooks, cell phones, MP3 players and so on.
Take a quick look at the table, it shows a list of benchmarks that are commonly used for various hardware components and portable gadgets. You'll notice that there are benchmarks which are common to all hardware and some are specificallydesigned for a few. In this story we shall take you through them and explain what each benchmark is intended for.
A benchmark can be classified as either synthetic or real-world. As the name suggests, synthetic benchmarks are applications that emulate real-world scenarios to measure the performance of a hardware component. Usually, they come in the form of suites that are a collection of individual benchmarks that target specific hardware components. Some suites such as in PC Mark and 3DMark, takeinstructions at one go so that all the benchmarksrun consecutively without human intervention. Some suites require each benchmark to be executed individually, such as SiSoft Sandra.
A sequence of benchmarks usually includes tests that evaluate the harddrive, graphics card, processor, RAM and so on. Most suites have preset scores for a plethora of hardware, which are used as references for rating the hardware which is being tested. Therefore, suites do not require much configuration.
In contrast to synthetic benchmarks, most real-world benchmarks are actual applications used to perform real-world tasks. Also they need to be configured and executed manually, as opposed to automatically. This is because a benchmark needs to be tweaked according to the hardware being tested. (Note: Settings may change across categories but within a category they are constant.) For instance, since a powerful benchmark cannot be used for entry-level hardware, the benchmark is tuned to match the hardware's capabilities.
There are a number of real-world benchmarks. For instance, to benchmark a processor, one can use video transcoding and audio encoding software. We shall learn more about benchmarking patterns as we proceed.