Mining Efficiency, J/TH
Miner efficiency is energy spent per unit of work. For SHA-256 it is measured in joules per terahash: watts divided by TH/s.
What efficiency and J/TH mean
Take the miner's power draw in watts and divide it by hashrate. Antminer S21: 3500 W over 200 TH/s gives 17.5 J/TH. That is efficiency: how many joules of energy the asic burns to compute one terahash. The lower the number, the better.
Why this number rules. Revenue per terahash is the same for everyone, set by hashprice. The electricity cost of a terahash, though, is different for everyone, and it is set precisely by J/TH multiplied by your rate. Two asics with equal hashrate and different efficiency are two different businesses.
The spread in a live catalog is enormous. The best asic among our 212 models is the SealMiner A4 Ultra Hydro at 9.45 J/TH, while the worst veterans sit around 157 J/TH: almost a seventeenfold gap. The line between life and death runs between them: as difficulty grows, the right edge of that scale goes dark first.
And the mandatory note on units. J/TH applies to SHA-256 and kHeavyHash. Scrypt uses J/MH, X11 uses J/GH, Equihash uses J/kSol, RandomX uses J/kH. Numbers from different columns cannot be added or compared, the work is different.
Quick reference table
| Formula | watts divided by hashrate |
| Unit for SHA-256 | J/TH, joules per terahash |
| Other algorithms | J/MH for Scrypt, J/GH for X11, J/kSol for Equihash |
| Best in the catalog | 9.45 J/TH, a hydro model |
| Catalog spread | 9.45 to 157.53 J/TH on SHA-256 |
How to read this number when choosing
First, compute your threshold. A rough estimate for SHA-256: the maximum viable price per kilowatt-hour in euros is roughly 1.08 divided by the miner's J/TH. An asic at 17.5 J/TH has a threshold around six cents, one at 30 J/TH around three and a half. Compare that with your rate before you even look at the ASIC's price tag.
Then remember that spec-sheet efficiency holds under spec-sheet conditions. Heat, dust on the heatsinks, and sagging voltage make real J/TH worse than factory. Hydro and immersion hold the number steadier than air precisely because their chips don't overheat in summer.
Firmware moves efficiency in both directions too. Undervolting lowers J/TH at the cost of some hashrate, while overclocking almost always makes it worse: consumption grows faster than output. On expensive electricity, a thrifty profile earns more than an overclocked one.
Example on real asics
S21 and WhatsMiner M66 side by side: 17.5 versus 19.9 J/TH. On paper the difference is modest, a bit over two joules. At the Spanish rate of 0.135 euros and around-the-clock operation, that means different power bills for every satoshi mined, and over a year it adds up to a sum comparable to a asic's monthly bill. Efficiency is the money you pay silently.
Related terms
Where to go next on the site
Hardware
- Antminer S21 tool
- WhatsMiner M66 tool
- ASIC Chip and Hashboard knowledge base
Algorithms
Numbers
Questions and answers
What J/TH counts as good today
Recent SHA-256 generations live in the 9.45 to 20 J/TH range. Anything above thirty, at European rates, deserves a calculator in hand, most likely the economics won't add up.
Why hydro versions are more efficient
Water keeps the chips in a narrow temperature corridor, so leakage and degradation are lower. Plus the case has no fans, which used to eat watts too.
Does J/TH change over time
It gets worse: chips degrade, heatsinks clog, thermal interface ages. Regular cleaning and a sane room temperature slow that drift down.
Can you compare J/TH with J/MH
No. These are different algorithms and different work. Compare asics only within one algorithm and one unit.
How the terms connect
Every link in the chain is clickable. Orange marks where you are now.
Looking for an ASIC miner
The catalog holds 212 models. You can compare them by hashrate and by joules per terahash, then plug your own rate into the calculator and see what stays in your pocket.
Page written and checked by Denys Klimchuk. Updated .