ITPEC FE Morning October 2018 Question 15

Running a standard program chosen for the intended use and measuring performance — that is a benchmark test.

A benchmark runs the same standard program, picked to resemble the intended workload (for example SPEC CPU for computation or TPC for transaction processing), on each system and measures the processing performance. Because the program is fixed, the results compare systems fairly on the kind of work they will do.

Answer (c)

Why not others:

• •(a) — estimating time from instruction frequencies and execution counts, without running anything, is an analytical method such as the instruction mix
• •(b) — building a model and running it with future estimates is simulation
• •(d) — measuring a running system's resource use and response with a monitoring program is performance monitoring

Key rule: A benchmark executes a real, standard program and measures it; instruction mixes estimate on paper, simulation runs a model, and monitoring observes live operation.

System Component · AI-written, may contain errors