There is no single or industry-mandated measure of the performance of a computer system or its constituent parts, and different uses of the product can impact performance. Benchmarks and performance tests (“benchmarks”) measure different aspects of the overall performance of computer systems and components using protocols designed to help simulate consumer use. For example, some benchmarks measure system performance on specific tasks using commercially available software applications, such as editing a photo or video; other benchmarks use software programs created specifically to measure the performance of a system or its components.
Benchmarks reflect many judgments about system configurations, workloads, and measurement methodology as to which reasonable benchmark developers may make different judgments, which may affect the results. Ultimately, benchmarks are intended to reflect how consumers may use products, but the actual performance any user may experience may be significantly different than the performance as measured by one or more benchmarks.
Benchmarks fall into two basic categories: (1) component benchmarks and (2) system benchmarks. "Component benchmarks" measure the performance of specific parts of a computer system, such as a microprocessor or hard disk drive. "System benchmarks" measure the performance of the entire computer system, including all of the components. In both cases, the performance seen in day-to-day usage generally varies from published benchmark performance. This is true for several reasons. First, components must be tested in a complete computer system, and it is not always possible to eliminate the considerable effects that differences in system design and configuration (such as the type of hard drive or amount of memory) have on benchmark results. Second, differences in software applications, operating systems, and compilers (programs used to generate executable software for a given platform) affect performance – these differences are unavoidable and will vary significantly across consumers and devices. Finally, benchmark tests are typically written to simulate the usage of certain types of computer applications, which may or may not be similar to the way that a particular user actually uses the application.
No single numerical measurement can completely describe the performance of a complex device like a microprocessor or a personal computer, but benchmarks can be useful tools for comparing components and systems. Nevertheless, the most accurate way to measure the performance of your computer system is to test the actual software applications that you use on your own system. The benchmark results published by Intel may be inapplicable to your component or system or your specific use case.
Benchmarks are only one kind of information that you may use during the purchasing process. To get a full picture of the performance of a component or system you are considering, you should consult other sources of information (such as performance information on the exact system you are considering purchasing). If you have questions about the performance of any Intel® microprocessor, please view the detailed performance briefs and reports published by Intel.
Intel contributes to the development of benchmarks in various ways. Intel is a member of or participant in various benchmarking organizations and consortia such as BAPCo* and SPEC*, and its employees often serve in various leadership roles. Intel also contributes programming resources, technical support, and/or funding to groups that develop benchmarks.
Intel is also a sponsor and member of the BenchmarkXPRT* Development Community, and was the major developer of the XPRT* family of benchmarks. Principled Technologies is the publisher of the XPRT* family of benchmarks. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases.
Measuring New Intel® Core™ Processors at Launch
Intel is a component supplier – our customers use Intel® processors and components to make amazing products for consumers after we launch these products. Because we cannot test the actual consumer-facing products that incorporate our components before launch, we often measure, estimate or simulate performance using an "Intel Reference Platform", or by using architecture simulation or modeling. An Intel Reference Platform is intended to be similar to, although not exactly the same, as the products we expect our customers will create with our components. The performance of Intel® processors on the Intel Reference Platform may be different from actual systems (once available).