Benchmarks

Benchmarks are an attempt to predict performance of a processor. They are updated every few years as computer specifications improve. They measure everything from simulation of desktop graphics programs to battery life.

Benchmarks are programs that are are written in a high level language that are designed to test the performance of a CPU.

Each benchmark program test a particular style of task:

systems,
numerical,
commercial

Using benchmarks makes measured performance data easily available so that a user can compare performace of different systems in handling particular types of task.

They are widely distributed and universally used by chip manufacturers to illustrate the strengths of their processors.

E.g. System Performance Evaluation Corporation (SPEC) —CPU2006

SPEC designed CPU2006 to provide a comparative measure of computer-intensive performance across the widest practical range of hardware using workloads developed from real user applications.

—CPU2006 for computation bound

– 17 floating point programs in C, C++, Fortran
– 12 integer programs in C, C++
– 3 million lines of code

—Speed and rate metrics

– Single task and throughput

SPEC produce benchmarks especially designed for graphics applications as these use a different mix of command instructions to numeric data handling programs. A processor that excels at one type of task may be only mediocre at another.