MINING TUNING AND OPTIMIZATION LAB

Deep Tune

Your miner tunes itself, on real mining, with real proof.

What it is

Every machine is different: chips, cores, memory, cooling. The settings that make one GPU fly can slow another down. Deep Tune finds your machine's fastest settings by testing them the only way that really counts, on your actual mining, against the real pool, for hours.

How it works

Open the Mining Tuning and Optimization Lab, the flask button on the mining card in easyBTX. Deep Tune lives inside it.
Pick a session length, 3 to 12 hours, and press Start. You keep mining the whole time.
Deep Tune changes one setting at a time in roughly 30-minute windows and measures how many difficulty-weighted shares your machine really produces. The pool automatically adjusts difficulty as you speed up, so Deep Tune scores the weighted rate, a number difficulty changes cannot fake or hide.
At the end it compares its best candidate against your baseline and applies a change only when the result is proven, which here means their 90% confidence intervals do not overlap. The gain is shown plainly, like “+23% vs your baseline”. If there is no clear winner it says so and changes nothing. That result still maps your machine.

What you see afterwards

A green “Deep Tune results arrived” pill appears on the mining animation when a session finishes. The panel always shows which settings are active right now, Standard or your fastest found, and switching between them is one click each way. Your fastest is remembered, even if you reset to Standard.

Sharing makes everyone faster

Every session writes a report into easyBTX's own data folder: your hardware, every window's settings and results, and your public worker and payout address so results can be checked against the pool. Never any keys, and never your seed phrase. A checkbox in the Lab, on by default, sends that report to easybtx.com so the shipped defaults can improve for every machine. You can untick it and share nothing. Separately, the moment a session ends a copy lands on your Desktop as a small .txt file, ready to post in the @easyBTX Telegram channel, and a button in the Lab saves it again if you ever lose that copy.

Where it runs today

Deep Tune is a Mac capability today. It scores accepted, difficulty-weighted shares reported by easyBTX's built-in engine through its pool connection. On Linux and Windows, mining after the MatMul v4.7 fork runs through the MATADOR engine, which does not report into that loop, so Deep Tune has nothing to measure there. We are not claiming it works on Linux or Windows until it does.

What it cannot do

Mining is statistical, so short sessions on slow hardware can end in “no significant winner”. That is the math being truthful, not a failure. Longer sessions give stronger answers. Deep Tune cannot make a slow machine fast, it can only find the best settings the machine already has, and on some machines the honest answer is that the defaults were already the best. It runs at 100% mining power during a session, because results have to measure your machine's true ceiling, and it restores your power setting afterwards. The disclosed 4.99% tool fee is unchanged, and tuning never touches it.

Deep Tune ships with easyBTX for Mac, on Apple silicon running macOS 26.4 or later. Get easyBTX →