Presets and J/TH: how efficiency actually raises margin
Undervolting cuts watts faster than hashrate, so the power bill drops more than output - that is the margin.
The efficiency preset lowers chip voltage: on an S21 Pro, power drops about 16% while hashrate drops only about 4%.
Energy per terahash (J/TH) falls from 15.0 to 13.1. At $0.095/kWh that lifts daily margin by more than 15%.
The maximum preset does the opposite - more hashrate for peak-price hours. Rollback to stock is always one click away, so testing costs nothing.
One preset serves one goal. Max Output raises hashrate at higher power draw, Efficiency lowers J/TH so power drops while hashrate may ease, and Power Target holds the wall-power level you set. Advertised gains describe different modes, never one simultaneous result, and the exact effect depends on silicon quality, cooling and ambient temperature.
How to compare two power modes on one Antminer
Compare modes A and B on the same miner with the same firmware version. Record the following for each mode, using comparable operating conditions and observation windows.
- Device and build: miner identifier, full model name, control-board platform and installed firmware version. The firmware catalog identifies model and build records; it does not provide measurements for your miner.
- Mode: the preset name shown on your device and the settings you changed. Keep the remaining settings the same so the comparison concerns the selected mode.
- Observation window: start and end time for each run. Compare equally long periods under comparable cooling and ambient conditions.
- Power: average watts, the measurement source and what it includes. Use the same measurement boundary for A and B; do not mix a device estimate with a wall-meter reading.
- Hashrate: average TH/s, the data source and its averaging window. Use the same source and observation window for both modes.
- Conditions and stability: ambient conditions, miner temperature range, cooling settings, errors and restarts during each run.
When hashrate is expressed in TH/s, calculate J/TH = average power in W / average hashrate in TH/s over the same window. Keep both original readings alongside J/TH: a lower ratio alone does not show how much hashrate was delivered or whether operation stayed stable.
Compare one device first, then follow the staged fleet rollout for matching model and board groups. Use the recorded conditions and results to decide whether to extend the trial.
Independent VNISH field reports
Follow a source-linked timeline of product results, models, watts, TH/s and operator observations from 2022 onward.
Install firmware
How to install firmware ModelsWork through the details
Open the practical guide for this specific situation.