A manufacturer rates an ASIC at 110 TH/s. You plug it in and your pool dashboard shows 108 TH/s on a good hour. By the third night, it drifts to 101 TH/s and your payout curve flattens. The unit is not dead — it is unstable.
Peak vs Sustained
Peak hash rate is measured during a short factory burn-in, often at ideal temperature and voltage. Sustained hash rate is what your unit delivers across days of continuous operation, including fan wear, dust accumulation, and ambient temperature changes.
For equipment rated at 0.05 TH/s and above, even a 3% variance compounds across a fleet. Ten units each drifting 3% below spec is effectively losing one unit’s worth of output.
What Causes Variance
Common contributors we see in the lab:
- Thermal throttling — Intake blockage or high ambient temperature forces the controller to reduce clock speed.
- Power supply sag — Cheap PSUs deliver lower voltage under sustained load, reducing hash board efficiency.
- Fan degradation — Bearing wear increases operating temperature over weeks, not hours.
- Firmware batch differences — Two units of the same model can ship with different controller firmware revisions.
How to Read a Stability Chart
Look for the floor, not the ceiling. A good unit shows a flat line with minor oscillation (typically under 2% for modern ASICs). A problematic unit shows downward drift after the 24-hour mark, often correlating with rising chip temperature.
When to Test
Before scaling a purchase. Before accepting a used lot. After any repair involving hash boards or thermal paste. And after moving equipment to a new room with different airflow.
Our 72-hour assessment is designed to surface these patterns before they become fleet-wide problems.