Glossary/Thermal/Thermal decay
Thermal decay
Thermal decay is the measurable fall in performance from the start of a sustained load to its steady state as the system heats and hits its limits.
Also known as performance decay, sustained-load decay
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.
Express it as a percentage: how far throughput at the end of a run sits below throughput at the start of the same unchanging workload. Because the work never changes, any drop is the system responding to its own accumulated heat and power state: heat, a power limit, or a current limit.
Decay is the number that tells you what your machine is like to live with. A system with 3% decay behaves the same in minute one and minute thirty. A system with 30% decay is effectively a different, slower computer once it warms up, and no peak benchmark score will tell you that.
This is the metric SystemCheck is built around. The CPU stage measures decay across a 60-second sustained multi-core load, and both GPU stages measure decay across their runs. It is directly measurable in a browser because it is derived from timing alone, unlike temperature or power. What SystemCheck cannot do is tell you which limit caused the decay. Pair it with a hardware monitor for attribution.
Related terms