VNISH NINJA · operator field notes · source linked

What operators noticed after tuning

These are practical signals from real machines: what the operator changed, what the miner did next and which detail must be checked before anyone copies the setup.

Published 28 August 2026

6source-linked reports

2022-2026years of field history

3S19 · S19j Pro · T21

Six field stories, read as operating scenarios

27 May 2026Recent field signal

Featured community result

Antminer S19 95T

Solar S19 profile at 1,100 W and 15.9 J/TH

A solar operator reported an S19 95T at 1,100 W and 15.9 J/TH, while praising the interface, autotuner and per-chip tuning.

Source-reported1,100 W · 15.9 J/TH

Derived from reported valuesimplied ≈69.2 TH/s

The captured post does not state whether the figures came from wall measurement, the controller, or an average over a stated time window. Treat 15.9 J/TH as an unverified source report.

Operator takeaway: copy no preset blindly; log model, board, power ceiling and cooling, then test one canary.

SolarEfficiencyTuning

RedditPublic independent-user contributionOpen original source ↗

25 October 2025Recent field signal

Featured community result

Antminer, model not stated

Granular power control and a modern interface

An operator highlighted fine power control and the interface, and reported continued operation during a fan-speed reporting fault while temperature remained the key protection signal.

Source-reportedPower control · interface · sensor context

Operator takeaway: inspect the underlying sensor and error state before changing a protection rule.

Power controlTuningStability

RedditPublic independent-user contributionOpen original source ↗

22 February 2025Recent field signal

Featured community result

Loki-modded Antminer S19j Pro

Two low-power points on a Loki-modded S19j Pro

The operator reported about 45 TH/s at 1,200 W and 28 TH/s at 900 W on a Loki-modded, two-board S19j Pro using a 120 V APW3 supply.

Source-reported45 TH/s / 1,200 W · 28 TH/s / 900 W

Derived from reported values≈26.7 J/TH · ≈32.1 J/TH

Operator takeaway: copy no preset blindly; log model, board, power ceiling and cooling, then test one canary.

EfficiencyPower control

RedditPublic independent-user contributionOpen original source ↗

4 June 2024Recent field signal

Featured community result

Antminer T21

A T21 set to a 175 TH/s target

A forum operator reported a first T21 running VNISH at a 175 TH/s target in a single-phase conversion using backup S19j Pro power supplies.

Source-reportedT21 · target 175 TH/s · single phase

Operator takeaway: copy no preset blindly; log model, board, power ceiling and cooling, then test one canary.

TuningPower controlStability

BitcointalkPublic independent-user contributionOpen original source ↗

21 September 2022Historical field signal

Featured community result

Antminer S19j Pro 104T · oil-immersion fleet

Immersion fleet: two S19j Pro 104T operating points

An operator working with an oil-immersion S19j Pro 104T fleet reported about 128 TH/s at a 4,300 W VNISH setting and around 124 TH/s at roughly 3,800 W, noting the efficiency trade-off.

Source-reported128 TH/s / 4,300 W · 124 TH/s / 3,800 W

Derived from reported values≈33.6 J/TH · ≈30.6 J/TH

Operator takeaway: copy no preset blindly; log model, board, power ceiling and cooling, then test one canary.

TuningPower controlEfficiency

RedditPublic independent-user contributionOpen original source ↗

18 June 2022Historical field signal

Featured community result

Antminer S19 family

Undervolting used to contain heat

In an S19 undervolting discussion, an operator described using VNISH for a strong underclock specifically to limit heat.

Source-reportedThermal objective · underclock

Operator takeaway: define the thermal limit first, then tune below it and keep a rollback path ready.

Efficiency

RedditPublic independent-user contributionOpen original source ↗

Source policy

Turn every story into a canary

A useful field report is not a preset. Identify the exact machine and board, set a power ceiling, prepare recovery, then validate temperatures, errors and pool-accepted work on one miner before scaling.

Prepare the install and recovery route →

How this collection is built

We include product-specific positive field reports with a dated public original and a direct source link. Distributor, reseller, hosting-provider, affiliate and sales-led contributions are excluded. The platform, date and permalink remain visible. Account handles are withheld unless the author gives explicit permission.

Hardware condition, board, power supply, cooling, ambient temperature, pool measurement and observation window can change the result. Treat every number as a lead for a controlled canary, not a promise.

The five gold stars are a visual collection marker, not a numeric rating assigned by the author.

Machine-readable evidence · JSON Schema.org · JSON-LD