Skip to content
Small team, full backlog, zero orders dropped. Support replies are slower than we’d like. Read our status update → Zero orders dropped. Status → 📬 Check your spam folder — most of our replies land there. We do answer. Status update → 📬 Check your spam folder. Status →

Antminer S19k Pro Underclock

The honest answer to “S19k Pro underclock settings”: pick any watt target from 2,050 W up — the curve is flat. Normal-mode efficiency holds 25.6–26.0 J/TH across the whole 2,050–4,860 W envelope, so your target is a heat, noise and circuit decision, not an efficiency trade. There is no frequency or voltage to copy: the autotuner derives both for your unit. The S19k Pro is also one of only two models here cross-checked on the D-Central bench.

Quick answer

The Antminer S19k Pro tunes across a 2,050–4,860 W envelope (Normal mode: 80–188 TH/s). Best catalog efficiency is 25.6 J/TH, and the sweet spot — everything within 5% of it — spans 2,050–4,860 W. The floor is 2,050 W — 74% of the 2,760 W factory rating. There are no frequency/voltage settings to copy: you set a watt target and the firmware autotuner derives the rest for your unit. Status: Verified (measured).

The efficiency curve is essentially flat across the envelope, so the watt target is a heat, noise and circuit decision — underclocking costs almost nothing per terahash.

Antminer S19k Pro — Normal mode Verified (measured)

2,050 W – 4,860 Wwatt envelope
80 TH/s – 188 TH/shashrate span
25.6 J/THbest efficiency @ 2,050 W–2,180 W
2,050 W – 4,860 Wsweet spot (≤5% off best)
Flatcurve shape
22catalog points

Floor exceeds both 120 V budgets (1,440 / 1,920 W) — this is 200–240 V territory. Budgets are the NEC 80% continuous-load rule; circuit fit only — stock Bitmain APW-series PSUs require 200–240 V input, so any 120 V build needs a PSU change regardless of watt target. At the floor this mode dissipates about 6,995 BTU/h (watts × 3.412).

Antminer S19k Pro — Performance mode Verified (measured)

2,850 W – 5,000 Wwatt envelope
78 TH/s – 135 TH/shashrate span
34.8 J/THbest efficiency @ 3,655 W
2,850 W – 4,750 Wsweet spot (≤5% off best)
Knee (high)curve shape
12catalog points

Floor exceeds both 120 V budgets (1,440 / 1,920 W) — this is 200–240 V territory. Budgets are the NEC 80% continuous-load rule; circuit fit only — stock Bitmain APW-series PSUs require 200–240 V input, so any 120 V build needs a PSU change regardless of watt target. At the floor this mode dissipates about 9,724 BTU/h (watts × 3.412).

Source: Firmware-extraction set, cross-checked on the D-Central live test unit (the DCENT_OS S19-generation development miner). The only S19-generation model here with a measured cross-check on both modes.

How to set a power target on this model

Aftermarket firmware exposes underclocking as a power target (sometimes "power limit" or a preset picker), not as frequency/voltage fields. Pick a watt figure inside the envelope above; the autotuner benchmarks your boards and settles onto it over a tuning period. Braiins OS+, VNish and LuxOS each expose this in their own UI, and DCENT_OS uses watt-target autotuning on its supported models; the undervolting guide walks through the firmware side step by step, and the firmware comparison covers which firmware supports which model. Verify the result at the wall with a meter — the target is an input, the wall draw is the fact.

Frequency and voltage are calculated at runtime by the firmware autotuner — never fixed presets; per-unit silicon quality, ambient temperature and cooling shift real results.

These envelopes are derived aggregates (bounds, bands, counts) — the row-level operating points, the interactive picker and the CSV/JSON of the source data live in the ASIC Power Profiles Database, the single source dataset for every number on this page.