Rejected Shares
Rejected shares are submitted solutions that the pool did not credit. The asic computed them and burned electricity on them, and no money comes back.
What rejected shares are
A pool accepts a solution only when it arrives in time, matches the current job and reaches the set threshold. Everything else lands in the rejected pile. A small share of such loss is always there, and it is the normal price of having a network between your asic and the pool.
The causes fall into four clear groups. Stale: the solution arrived after the pool changed the job, and the latency to the server is to blame. Duplicates: the same solution was sent twice. Below the threshold. And invalid, which most often means chips running past the edge of stability.
Half a percent is normal. One or two percent is tolerable, though it is already worth a look at the network. More than three percent is a symptom, and you take it apart in order: latency to the pool first, then the cable and the switch, then temperature, and only then overclocking.
One case we see in the shop regularly: aggressive overclocking lifts hashrate by a few percent and lifts rejects from half a percent to five at the same time. Revenue drops, because the pool pays for accepted solutions and not for computed ones.
Quick reference table
| What it is | solutions the pool did not credit |
| Normal | up to half a percent |
| Reason to check the network | one to three percent |
| Sign of trouble | more than three percent |
| Main causes | latency to the pool, network, heat, overclocking |
How to work through it in order
Start with geography: a pool server in Europe rather than across the ocean. Every extra hundred milliseconds of latency turns into stale solutions.
Next comes the physical network. A cheap patch cord, a crushed connector or a lossy switch port produce a steady stream of rejects that looks like a fault in the miner. A new cable costs less than an hour of diagnostics.
Then temperature and the firmware profile. If rejects climb along with the room temperature, the problem is cooling. If they appeared right after a profile change, roll the profile back and compare: calm settings often bring in more money than overclocked ones.
An example in numbers
An Antminer S21 at 200 TH/s with half a percent rejected loses about one terahash of useful work. At five percent the loss is already ten terahashes, so the asic performs like a 190 TH/s model while the meter still runs for the full 3500 W. Overclocking that added three percent of hashrate and five percent of rejects has put you in the red.
Related terms
Where to go next on the site
Hardware
- Antminer S21 tool
- ASIC Miner knowledge base
Software
- Software firmware
Questions and answers
Do rejected shares get paid
No. The pool counts accepted ones only, so every extra percent of rejects is a direct loss of money against the same power bill.
What is a stale share
A solution that arrived after the job changed. Usually the latency to the pool server is to blame, less often an unstable network inside the building.
Can the firmware be at fault
Yes, when the profile pushes the chips past the edge of stability. The sign is simple: rejects grew right after a profile change while temperature and network stayed the same.
How many rejects does a healthy asic have
Up to half a percent over a day. A five minute window tells you nothing, because the spread there is normal.
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 .