Glossary/Benchmark/Synthetic vs real-world benchmark
Synthetic vs real-world benchmark
A synthetic benchmark runs a controlled artificial workload for repeatable comparison; a real-world benchmark times actual application work you care about.
Also known as synthetic benchmark, real-world benchmark, application benchmark
Measured by SystemCheck
Part of the four-stage run: a 60-second sustained CPU load reporting throughput, decay and consistency; a memory stage reporting bandwidth and latency; a calibrated GPU raster load reporting frame time, pacing and decay; and a path tracer reporting samples per second. The paragraphs below say which metric carries this term.
Synthetic tests exist to be repeatable. They fix the workload so that run-to-run and machine-to-machine differences reflect hardware rather than scene complexity, asset streaming, or patch version. That control is what makes them useful for isolating a variable (did this cooler change help?) and for detecting regressions.
Real-world tests measure what you actually do: this game at these settings, this project compiled, this timeline exported. They are the ground truth for a purchase decision but are noisy, hard to reproduce exactly, and go stale as software updates.
The correct use is complementary. Use a synthetic test to check whether the machine is healthy and behaving consistently, and a real-world test to decide whether the machine is fast enough for your work. SystemCheck is deliberately synthetic: fixed workloads, fixed durations, decay and consistency reported. It answers "is this system holding up under sustained load", not "how many fps will I get in a given game".
Related terms